The Freer Chinese bronzes

The author and the Director of the Gallery examine a Chinese bronze in the laboratory.

SMITHSONIAN INSTITUTION

FREER GALLERY OF ARTORIENTAL STUDIES, No. 7

THE FREERCHINESE BRONZES

Volume II Technical Studies

BY

RUTHERFORD JOHN GETTENS

WASHINGTON1969

(SMITHSONIAN PUBLICATION 4706)

PRINTED BY THE MERIDEN GRAVURE COMPANY, MERIDEN, CONNECTICUT

ORIENTAL STUDIES SERIES

No. 1. The Story of Kalaka, 1933

W. Norman Brown

No. 2. A Descriptive and Illustrated Catalogue of MiniaturePaintings of the Jaina Kalpasutra, 1934

W. Norman Brown

No. 3. A Descriptive and Illustrative Catalogue of Chinese

Bronzes Acquired During thf Administration of JohnEllerton Lodge, 1946

Lodge, Wenley and Pope

No. 4. Shiraz Painting in the Sixteenth Century, 1949

Grace Dunham Guest

No. 5. Thf Field of Stones, 1962

Richard Edwards

No. 6. Armenian Manuscripts in the Freer Gallery of Art, 1963

Sirarpie Der Nersessian

No. 7. The Freer Chinese BronzesVolume I, Catalogue, 1967

Pope, Gettens, Cahi/I and Barnard

Volume II, Technical Studies, 1969

RutherfordJohn Gettens

IN MEMORIAMARCHIBALD GIBSON WENLEY

(1898-1962)

^BRAS!^7 CONTENTS

Chapter Page

LIST OF TABLES ..... xi

FOREWORD ..... xiii

PREFACE ...... xv

I BRONZE: CONSTITUENTS AND BEGINNINGS . 1

II BRONZE IN ANCIENT CHINA . . . 13

III COMPOSITION ..... 33

IV FABRICATION ..... 57

V METALLOGRAPHIC STRUCTURE . . 121

VI INSCRIPTIONS ..... 141

VII RADIOGRAPHY . .159VIII PATINA AND CORROSION ... 171

IX FILLINGS, INLAYS AND INCISED DECOR 197

X FALSE PATINA AND REPAIRS ... 209

APPENDIX I ..... 229

The validity of the analytical data

APPENDIX II ..... 239

Published analyses of illustrated bronzes

BIBLIOGRAPHY ..... 243

INDEX ..... 253

ix

LIST OF TABLES

Page

Table IA MAJOR CONSTITUENTS ... 48

Table IB MINOR AND TRACE CONSTITUENTS . 49

Tables II—V COMPOSITION IN RELATION TOCHRONOLOGICAL SEQUENCE . . 54

Table VI ANALYSIS of TING SC587 ... 80

Table VII ANALYSIS OF TING ROM.933.23.64 . . 82

Table VIII ANALYSIS OF CHAPLETS IN KUVESSELS . . . . .107

Table IX CORE MATERIALS: CLASSIFICATION . Ill

Table X CORE MATERIALS: PARTICLE SIZES . Ill

TableXI ANALYSIS OF KUEI 24.11 ... 138

Table XII ANALYSIS OF SOFT SOLDERS . . 212

Tables A to E APPENDIX I . . . . .230Table - Appendix II, published analyses of illustrated bronzes . 239

xi

FOREWORD

This volume of Technical Studies supplements the brief paragraphs

headed Technical Observations that appear under each bronze in Volume [,

The Catalogue. Together they embody the results of the sustained and in-

tensive research program carried on by Rutherford John Gettens, HeadCurator of the Freer Technical Laboratory, and his staff over the last ten

years and more. Acknowledgement of the help he has received from

members of the Freer staff and from outside collaborators is made in the

author's preface. It is appropriate here to record the special contribution of

W. T. Chase who first took part in the investigations as an intern graduate

student from the New York University Conservation Center of the Institute

of Fine Arts, who later joined the regular staff in the final year of the study,

and who assumed direction of the Laboratory on Mr. Gettens' retirement this

year. In addition to his contributions in the scientific field, he has been most

helpful in the editorial work, and during the author's illness, he displayed

efficiency and confidence in collaborating with the Editor in preparing the

manuscript for the press.

To the best of our knowledge, this is the first time a volume has been

devoted to the detailed scientific analysis of a major collection of Chinese

ceremonial bronzes. We publish it in the hope that it may contribute to the

advancement of investigation into the methods and materials of ancient

craftsmen.

JOHN ALEXANDER POPE

Director

Freer Gallery of Art

Washington, D.C.

November, 1968

Xlll

PREFACE

This technical study on Chinese bronzes was first suggested by the late

Archibald Gibson Wenley, second director of the Freer Gallery of Art. It

was Wenley who in 1951 invited the writer to help him establish the Freer

Laboratory to study the materials and methods of ancient craftsmen in

Asia. Mr. Wenley had come to realize, during his many years of sinological

experience, that there was something lacking in art historical studies limited

purely to matters of style, chronology and epigraphy, and that knowledge

and understanding of the material aspects of antiquities was of equal im-

portance. He well understood that the objects of the Freer collection served

not only as material for the study of the evolution of ideas and of the artistic

expression of those ideas, but also that the history of technology in the

Orient was reflected in those treasures. I well remember the day he walked

into the laboratory and, after telling me that the Descriptive and illustrative

catalogue of Chinese bronzes by Lodge, Wenley and Pope, published in 1946

was out of print, suggested that the preparation of a new catalogue be con-

sidered. Since laboratory facilities were now available, he suggested that

the bronzes be subjected to technical study and the findings incorporated

along with the usual catalogue descriptions. He invited me to collaborate.

In the twenty-odd years that have intervened since the publication of the first

catalogue, many fine new bronzes have been added to the collection. It

was finally decided to limit the study to ceremonial vessels but to include all

kinds, inscribed, uninscribed, early and late, and even vessels of doubtful

pedigree. The laboratory studies began intensively in 1958. In June, 1961, a

small seminar on "Technical Studies of Ancient Metal Artifacts" was held

at the Freer in which the plans for study were outlined. Unfortunately,

Archibald Wenley did not live to see the project finished. When John

Alexander Pope assumed directorship of the Freer in 1962 he inherited also

the supervision of the bronze studies. Although, as is well known, his first

love is ceramics, he pitched into the bronze problem with zeal and en-

thusiasm and took on the sinological side of the studies and the manifold

details of catalogue preparation. He gave abundant support to the technical

studies, and it was he who suggested that the findings be summarized and

put together in a separate volume, closely tied to the catalogue. His help in

bringing to completion this task and text has been immeasurable.

These studies could not have been carried out by a single person. It

turned out eventually to be a cooperative undertaking of some size in which

xv

many specialists participated. First, the author would like to recall the

memory of the late Langdon Warner of Harvard University who introduced

him to Chinese art and imbued him with a special interest in antiquities of

metal. It was not hard for a chemist to fall under the spell of early Oriental

metallurgy. Warner would have watched the unfolding of evidence with the

keenest interest.

In the Freer Gallery many members of the staff have been involved.

Elisabeth West FitzHugh, before her retirement, did much of the analytical

work involving both wet methods of chemical analysis and spectrometry.

She also carried out most of the X-ray diffraction analyses on corrosion

products and the photomicrography of polished metal samples. Later

much of this work was taken over most capably by Mrs. Ilona Bene.

Although Mr. W. T. Chase joined the project rather late, he contributed

much in the way of special studies on metal structure and composition of

clay cores; he has also helped immeasurably in the preparation of the manu-script. Mrs. Meryl Johnson also assisted in the bench work by taking a

number of the photomicrographs used in the illustrations. Members of the

Freer staff outside the laboratory have given generous support to the pro-

ject. Mr. Martin Amt, Museum Specialist, answered innumerable calls for

help in bringing bronzes to the laboratory from storage or from the exhibi-

tion cases; he also made all weight and dimension measurements. Membersof the Photographic Laboratory, Mr. Raymond A. Schwartz, Chief, and

his assistants, Mr. James W. Riggs and Miss Ursula Pariser, worked side

by side with members of the Technical Laboratory making photographic

details and supplying prints. Their long-suffering patience throughout has

been admirable. Mrs. Priscilla Smith, Librarian, and Mrs. Aleita Hogen-

son, formerly Assistant Librarian, supplied invaluable help with the bibliog-

raphy; and Mr. Frank A. Haentschke painstakingly prepared drawings

and mounted all photographs. Not least have been the contributions of

Mrs. Daphne McCloskey, Miss Lucille Aldrich and Mrs. Elizabeth K.

Hamel who have cheerfully typed the manuscript through countless drafts.

Editor Lloyd Langford's meticulous work in preparing the manuscript for

printing warrants high praise.

Many persons outside the Freer have also contributed generously, and

all on a personal favor basis. Outstanding is the aid given by staff membersof the Non-destructive Analysis Group of the Naval Ordnance Laboratory,

White Oak, Silver Spring, Maryland, especially Mr. Daniel Polansky,

Mr. Edward L. Criscuolo and Mr. Donald Case. The several dozen radio-

graphs of bronzes made by this team using high energy X-ray equipment has

opened up new vistas in the study of metal antiquities. Mr. Charles

xvi

Bridgman of Eastman Kodak gave helpful advice on the reproduction of the

radiographs. Through the courtesy of Dr. George Switzer, Chairman of the

Department of Mineral Sciences, Smithsonian Institution, specialists of his

department have lent their talents to the problem: Mr. Grover Moreland

prepared numerous specimens of bronze for metallographic study and Dr.

William G. Melson helped in petrographic studies on earth materials of

cores and molds. Mrs. Jacqueline Olin of the Conservation-Analytical

Laboratory helped with statistical studies on analytical data and Mr.

Maurice Salmon made X-ray fluorescence analysis of numerous specimens.

Through Dr. Edward V. Sayre, facilities for spectrograph^ analysis of

metals were made available at the Brookhaven National Laboratory, Upton,

New York, and at the Consevation Center, Institute of Fine Arts,

New York University; and he gave much wise counsel during the course of

the studies. Greatly appreciated is the help given by colleagues at the Royal

Ontario Museum, Dr. Henry Trubner and the late Mr. William Todd, whomade available for study and comparison that outstanding collection of

early bronzes made in China by the late Bishop William Charles White.

Dr. Noel Barnard of the Australian National University, Canberra, during

two extended visits to Washington joined in technical examination of the

bronzes and made many pertinent observations on fabrication. Professor

Cyril S. Smith of the Massachusetts Institute of Technology was interested

in the project from the beginning and made several visits to the Freer to

participate in discussions and to give guidance. Professor Smith and his

assistant Mrs. Katharine Ruhl, who made metallographic studies on speci-

mens sent to Cambridge, contributed much to the substance of Chapter V.

The late Orvar Karlbeck of Lidingo, Sweden, who was enabled to visit the

Freer in 1964 through a grant from the American Council of Learned

Societies, contributed insight and information out of his years of experience

in the study of bronzes while working as a civil engineer in China. During

two delightful days in the spring of 1964, Pope, Karlbeck, Barnard, Smith,

and I gathered around bronzes set on the examining table in the Freer, and

the heat of discussion and debate raised a din that has scarcely died down.

Others have kindly read the manuscript and offered suggestions; Mr.

Theodore Wertime of the Voice of America read Chapters I and II dealing

with the history of metallurgy; and Professor Earle R. Caley of Ohio State

University read Chapter III on composition.

While most of the testing and experimental work reported here was done

directly on our own bronzes some of the most important of the technical

studies were carried out on vessels or fragments of no exhibition value given

to us by generous donors with permission to sample, cut up, dissolve,

xvii

destroy or to use in any way that would contribute to knowledge. Amongthese were three fragmentary ku and a p'an from Mr. Simon Kriger of

Washington, D.C.; a ting given by Mr. David Ketel of the Netherlands

Foreign Service while he was stationed in Washington; a ting received from

the Philadelphia Museum of Art through the kind offices of Miss Jean G.

Lee, Curator of Far Eastern Art; and finally a very important fragmentary

early Chou dynasty hu given some years ago to the writer by the late

Charles Fabens Kelley of the Art Institute of Chicago.

To all of these friends, collaborators, and donors the author owes a debt

of gratitude he can never repay.

RUTHERFORD JOHN GETTENS

Washington, D.C.

October, 1967

xviii

CHAPTER I

BRONZE: CONSTITUENTS AND BEGINNINGS

Bronze is the oldest artificial alloy. It is made of copper and tin, but it

often contains varying proportions of other elements lead, zinc and

occasionally silver. The origin of the name is obscure. Murray 1 says that it

became current in English usage during the 18th century when critical

examination of Greek and Roman metal antiquities showed that their com-position was different from that of the copper-zinc alloy then in use, called

"brass." The Italian word bronzo altered to "bronze" was borrowed to

denote this "ancient brass." Bronzo might have originated from the Latin

brundisium, modern Brindisi, a seaport of Italy famed in ancient times for

its bronze, or it might have come from the Latin brunitius, "brown colored."

Before the 18th century, "brass" meant all alloys of copper with tin, zinc or

other base metals; the term is used with this meaning in the King James

Bible, for instance. "Brass" came from the old English braes, which was

initially derived from a southwest Asiatic language; 2 braes probably has

the same source as the Hebrew and Phoenecian barzel, "iron." In modernusage, "bronze" denotes a reddish-yellow alloy of copper and tin, while

"brass" denotes a yellow alloy of copper and zinc. 3 In Chinese the character

fung* stands for copper, brass, or bronze; the exact meaning is supplied

by the context or by a modifier.

With the exception of copper, none of the bronze-making metals occurs

free in nature; and even copper, like tin and lead, most often occurs in

combination with other elements from which it must be freed to be useful.

There are no metallic ores, except possibly in rare and isolated instances,

which combine copper, tin, and lead in the correct proportion to produce

bronze by direct smelting. It is generally assumed, and there seems to be no

reason to deny it, that each of the essential metals of bronze was found as a

separate ore, that each ore was smelted by a process more or less peculiarly

suited to it, and that the bronze of the ancients was produced by the deliber-

ate combining of the necessary metals in predetermined proportions

although such predetermination may have been nothing more than the general

knowledge that they needed a lot of copper and a little tin. Archaeologists

new English dictionary on historical principles, Oxford and New York, 1888. Vol. I, p. 1 126.2See Philip Babcock Gove, editor, Webster's third new international dictionary ... p. 268.3Gove, op. cit. pp. 268 and 283.4 R. H. Mathews. Mathews' Chinese-English dictionary, p. 968, No. 6623.

i

generally agree that true bronze contains tin in excess of about 3 percent.

Coghlan points out that many prehistoric objects contain tin in amountsfrom 0.1 to 3 percent, and he terms these "accidental bronzes" in the

belief that they were made from copper ores in which tin was adven-

titiously included. 5 Many of the bronze weapons analyzed by Dono {vide

infra) seem to fall into this class.

Lucas 6 outlines four ways in which prehistoric bronze could have been

produced: (1) by fusing metallic copper and metallic tin; (2) by smelting a

mixture of copper ore with metallic tin; (3) by smelting a naturally occurring

combined mineral of copper and tin (stannite); or (4) by smelting either a

naturally occurring or artificially made mixture of copper ore and tin ore.

He goes on to say that the first two are out of the question unless tin was

known before copper, a situation for which there is no evidence. Stannite

reduction, because that mineral is rare in most parts of the world, is not

likely. Thus he concludes that early bronzes were deliberately made by

mixing the oxide and/or carbonate ores of copper and tin. Rickard concurs

in this view and relates the experiments of William Gowland, "who simu-

lated primitive practice by making a hole in the ground, in which he put a

mixture of ore containing 20 percent, of tin, as tinstone, the oxide, together

with an ore containing 30 percent, of copper, as malachite, the green car-

bonate. This mixed ore was placed in layers alternately with charcoal in the

hole, and when the fire was started a blast was applied through a one-inch

iron pipe. In this manner he made a bronze that contained 22 percent, of

tin, which would prove excessive for the primitive artificer, because an

alloy so rich in tin is too brittle for the making of tools, although well

adapted for the manufacture of bells." 7

Thomas T. Read is even more positive. He says 'There can be no doubt

that all the early bronze was made by smelting mixed copper-tin ores." 8 Hefurther reminds us that modern Monel metal, which is made from the

copper-nickel ores of Sudbury, Ontario, is also a natural alloy. The question

of methods employed in early smelting is very complicated as one can judge

from R. J. Forbes' review of the history of tin and the origin of bronze in

which he compares the views of various authorities on the history of

metallurgy.9 Forbes feels that the mixed oxides method of smelting mayhave been satisfactory for tin oxide in the form of "stream tin" but not so

good for "vein tin."

It should be mentioned that nearly all the theories on the discovery of5"Notes on the prehistoric metallurgy of copper and bronze . . .

," p. 23.

^Ancient Egyptian materials . . . (1962), p. 218.7"The primitive smelting of copper and bronze," p. 246.

^'Metallurgical fallacies . . .," p. 385.

^Metallurgy in antiquity, pp. 248-252.

2

bronze have been propounded by students of Europe or the Middle East

who have little or no knowledge of the Far East or of early metallurgical

attainments in that part of the world. They do not take into account the

presence of considerable proportions of lead in the copper-tin alloy which

is a characteristic of Chinese bronzes. The development of Chinese bronze

metallurgy, which we shall attempt to review later, may have taken a

different course from the very beginning.

As we shall see, the analytical results obtained from the Freer bronzes

seem to support the mixed oxides theory. The wide range of tin content

suggests that control of alloy composition was not good, and the product

is just about what we might expect from the reduction of ores of indefinite

and varied composition.

Before turning to a more detailed discussion of the three principal metals

that bronze contains, we may mention briefly the advantages of bronze over

pure copper. Even though pure copper melts at a moderate temperature, it is

a poor metal for casting because the melt is sluggish and viscous; it also

shrinks on cooling and does not sharply fill the details of the mold. It has

a tendency to absorb gas which causes porosity and blistering, but above all

to melt it required temperatures that must have taxed the ceramic materials

of early crucibles as well as the fire-making capacities of the early smelters.10

The addition of tin to copper increases the liquidity of the melt so that the

alloy reproduces more closely the details of the mold. The table below

shows that the fusibility of the mixture increases rapidly until a fairly high

content of tin is reached. 11 The same effects can be imparted by other im-

purities, like arsenic, but without the beneficial results of tin.

Melting points of copper-tin alloys

(After T. A. Rickard)

Copper: Tin Melting Point

100:0 1083°C

95:5 1040

90:10 994

85:15 944

80:20 875

75:25 795

10In this connection it is worth noting that the image makers of northern India and Nepal seem to have over-

come these difficulties in a way we do not understand, for they have produced many Hindu and Buddhistfigures of cast copper which reveal high technical as well as artistic quality.1 lSee also the constitution diagram for copper-tin alloys, figurel48, Chapter V.

3

COPPER

Sources

In spite of its extensive use in the modern world, copper is one of the less

abundant elements; according to Fleischer it accounts for only about 0.01

percent of the earth's crust. 12 The chief ores of copper are: native copper;

two sulfides with iron, chalcopyrite, CuFeS2, and bornite, CusFeS^ twooxides, cuprite, CU2O, and tenorite, CuO; and the two carbonates, mala-

chite, Cu2(OH) 2C03, and azurite, Cu 3(OH)2(C0 3 )2. The sulfides are the

primary ores and supply the great bulk of modern industrial copper, but

it has been generally held that they played little part in primitive copper

metallurgy because they are difficult to recognize and difficult to smelt.

This is the view held by Coghlan, but he thinks that sulfide ores were ex-

ploited during and after Roman times in Europe. 13 Caley and Easby found

evidence that the pre-Columbian Indians of Peru produced copper from

sulfide ores.14 Butts lists other ores (minerals) of copper such as sulfates,

silicates and arsenates, 15 but, because they are rare, they were never of

economic importance even to early metallurgists.

Copper ores available in antiquity

Most important in primitive metallurgy were the so-called "secondary

ores11

of copper which occur in the oxidized zones near the surface of ore

bodies. It is generally understood that these secondary copper ores are

formed from primary ores through the agencies of air and water. The pro-

cess of secondary ore formation is much like, and the products are identical

to, the formation of corrosion products on the surface of a buried bronze.

To the primitive metallurgist the most important of these secondary ores

were the oxides, cuprite and tenorite, the two carbonates, malachite and

azurite, and native copper.

The secondary ores of copper are important because, in addition to being

located near the ground surface, they also tend to have bright colors or

other characteristics by which they are easily recognized: malachite is

bright green; azurite, bright blue; cuprite is dull to bright red; tenorite is

black, but often the surface is weathered green to malachite; 16 native

copper is easily detected even when tarnished. These ores, moreover,

*-%U.S. Geological Survey Circular No. 285, p. 6.13"Notes on the prehistoric metallurgy of copper and bronze . . .

," p. 16.14'The smelting of sulfide ores . .

.

," p. 59.

^Copper: . . . , p. 22.16 In New Mexico the writer has seen a seam of green-stained earth some inches wide, which extends across the

countryside for perhaps a half-mile. In another place at an abandoned copper mine on a mountaintop, the

greenish color of the old ore dump could be seen from some miles distant. On closer inspection, the lumps ofslaty rock taken from the mountaintop mine were found to be traversed with seams of bright green brochantite

(basic copper sulfate).

4

are easily smelted with charcoal using simple hearths at moderate

temperatures.

Native copper

There has been much discussion about the role played by native copper

in primitive metallurgy, and no little divergence of views, but there is less

doubt of its importance because of recent archaeological discoveries in

Anatolia, Iran and Iraq. Wertime has called attention to finds of artifacts of

native copper in sixth and fifth millennium B.C. sites at £atal Hiiyiik

(Turkey), Chagar Bazar (Syria) and Sialk (north-west Iran.) 17 Native copper

is found in both primary and secondary ore deposits, and Clark says it is

thought to be derived from the reducing action of iron compounds on sol-

uble copper minerals. 18 It is often seen in microscopic quantities among the

corrosion products of ancient bronzes. Elemental copper is occasionally

found in stream beds as green or red corroded nuggets; and on the surface it

is sometimes found as small brownish boulders. Arborescent forms, which

are occasionally seen, are groups of small and often quite perfect metal

(cubic) crystals joined in branched strings and clusters. In some places

native copper is found in veins in trap rock. In a few localities, notably in the

Keweenaw Peninsula in northern Michigan, it occurs in large solid masses

weighing several hundreds of tons. It is well known that native copper from

northern Michigan was traded by the American Indians in pre-Columbian

times all over North America where it was used mostly for ceremonial and

ornamental purposes. Forbes says that native copper, which occurs in thin

plates and arborescent forms, is easily worked.19 Professor Cyril S. Smith,

who has recently discussed the working of native copper, adds that its

characteristic microstructure has been found in several objects of great

antiquity in Egypt, Asia Minor, Mesopotamia, Palestine and the Danubevalley.20

We can presume that many outcrops of native copper known to primitive

man have long since been exhausted and disappeared. Rickard says that

native copper is more abundantly distributed in nature than is generally

supposed.21 In fact, Thomas T. Read felt that copper metallurgy began

when early man tried to melt native copper to which copper minerals were

still adherent. 22 This is not an implausible theory.

17"Man's first encounters with metallurgy," p. 1258.18 77;e data ofgeochemistry, p. 672.19"Extracting...,"p. 585.20"Materials and the development of civilization and science," p. 908 ff.

21"The early use of the metals," p. 304.22"Metallurgical fallacies . .

.

," p. 383 ;quoted in Coghlan, "Notes on the prehistoric metallurgy of copper and

bronze . ..," p. 22.

5

Smelting ofcopper ores

The release of a metal from the active elements with which it is combined

is called "smelting." Smelting is a chemical reaction carried out at varying

temperatures in which oxygen, sulfur or other non-metallic elements or

groups of elements (i.e. sulfates) are removed. This is done by mixing the

finely crushed ore with carbon in one of its various forms, charcoal, coal, or

coke. The carbon at the bottom of the furnace is set on fire. The hot carbon

and the carbon monoxide gas from its combustion combine with the oxygen

of the copper oxide (cuprite or tenorite) or carbonate (malachite or azurite)

to form carbon dioxide which escapes as a gas to leave the molten metal free.

An interesting description of the primitive smelting of copper ore (mala-

chite) as it was observed earlier in this century (1911) by Monseigneur de

Hemptinne in the Katanga province of the Belgian Congo, is reported by

Rickard. 23 The smelting, which was performed by the Bayeke tribe of

Negroes, was done by mixing the green ore with charcoal in a small basin

dug in the earth. The charcoal is ignited, and the fire is fanned with a crude

bellows made from antelope skin; the air is conducted to the base of the fire

through pipes (tuyeres) made from termite cones. With the aid of rituals

performed by sorcerers, the founders, after several hours, find a pool of

copper at the bottom of the furnace. When cooled and freed from clinkers,

the crude ingot is ready for refining, which is later carried out in a second

furnace operation. According to Swarup and Misra, similar primitive cop-

per metallurgy was practiced until recent times in India. 24 In fact, primitive

smelting of copper was carried on in many widely separated places in the

ancient world where there is little chance that knowledge of that kind could

be gained by diffusion. 25 The entire question of the discovery of non-ferrous

metals in antiquity and the early metallurgy of copper and bronze has been

thoroughly explored by Coghlan and colleagues in England, but unfor-

tunately using mainly Western source materials. More recently Professor

Smith has brought into focus the relationship between properties of mate-

rials and development of civilization, using copper and its alloys as a striking

example. 26

Lucas, Rickard and Read27 feel there is good reason to believe that copper

and tin ores were smelted separately in the late Bronze Age and that the two

metals (with the possible addition of lead) were mixed to produce bronze.

23"The primitive smelting of copper and bronze," p. 233 ff.

24"Ancient copper industry ..."25The archaeological researches of Nelson Glueck and others have increased evidence that a primitive coppermetallurgy much like that described above, was carried on in the Negev of southern Israel in the second millen-

nium B.C. (See Illustrated London News, Sept. 3, 1960 and G. E. Wright, "More on King Solomon's Mines").26"Materials and the development of civilization and science."27 0/7. cit.

6

Ingots of raw and fairly pure copper are known from a fairly early date.

Copper ingots in shapes of discs and bars are known from the late Bronze

Age in the British Isles and Europe. In the Near East a strange flat rectangu-

lar copper ingot with projecting corners, called an "ox-hide'1

ingot, was

known at an early date. R. J. Forbes says that these probably served as

money in place of real ox hides used formerly.'28 Such ingots 29 are shown in

several Egyptian tomb paintings reproduced by Coghlan, R. J. Forbes and

others.30 Tylecote informs us that in England "bun" ingots of copper were

produced during the Roman occupation.31

Properties ofcopper

Pure copper, the result of the smelting process, melts at about 1083 C(1981.4°F), which is bright red heat. Its reddish color is unique amongmetals. It is malleable, ductile, and a good conductor of heat and electricity.

Pure copper solidified from the melt is soft and easily scratched with a

pocket knife. The metal can be hardened by mechanical action in the cold,

like drawing, rolling or hammering. The ancients in the West made practical

use of the hardening effect produced by hammering. Copper hardened by

cold working can be softened by reheating (annealing) to a temperature in

excess of 500°C. The metal can also be hardened by alloying with certain

other elements like tin, chromium, antimony and arsenic. Copper alloys

(i.e., forms solid solutions) easily with a few metals, notably tin, zinc,

nickel, silver and gold. It does not alloy well with either lead or iron.

Chemically, copper is moderately active, hence it is readily transformed

into copper compounds, many of which are brightly colored.

TIN

Sources

Even less abundant in nature than copper, tin, according to Fleischer,

makes up only about 0.004 to 0.0006 percent of the earth's crust. 32 Unlike

copper, tin occurs in a very limited number of ways, principally in the simple

mineral called cassiterite, which is stannic oxide, Sn(>2. Most authorities onearly metallurgy assume that this was the form in which tin was available

to primitive metal workers. Cassiterite, which has a theoretical tin content

of 78.6 percent, occurs mostly in the form of small brownish-black pebbles28"Extracting . .

.

," p. 587.29A number of ox-hide shaped ingots of crude copper were found under water in the wreck of a Bronze Ageship discovered in 1960 in the sea off Cape Gelidonya in southwest Turkey. The ship appears to date around1200 B.C., and the copper it carried was almost certainly from Cyprus. See George F. Bass "The Cape Geli-

donya wreck;" also Peter Throckmorton, "Oldest known shipwreck yields Bronze Age cargo."30"Notes on the prehistoric metallurgy of copper and bronze . . . ," p. 68; "Extracting . . . ," p. 578.

^Metallurgy in archaeology, pp. 29-34.^ 2 U.S. Geological Survey Circular No. 285, p. 5.

7

in alluvial areas and is commonly known as "stream tin.'1

The cassiterite

pebbles are very heavy (sp. gr. about 7); the mineral is stable, and Coghlan

informs us that it survives indefinitely in alluvial deposits.33 One other tin

ore called stannite should be mentioned; it is a complex sulfide of tin,

copper and iron, and has already been described as a rarity; even whenfound it is little used because it is difficult to reduce. It is said, however, to be

the principal ore in the Yunnan deposits of China, a matter which we shall

have occasion to discuss later.

On the whole very little is known about the early metallurgy of tin and in

most books on the subject it is overshadowed by copper. Aitchison, how-ever, gives a concise description of the smelting of cassiterite as follows:

"In the oldest practice this was achieved by filling a deep trench or pit with

small timber, igniting the wood and, when it was well ablaze, adding broken

tinstone and more wood alternately. The tin ran down to the bottom of the

fire. Yields were low, much ore being left unreduced and much tin being lost

in the slags and furnace ashes. Though technologically different, this pro-

cedure sufficiently resembled that used in lead smelting to encourage the

notion that tin was a variant of lead. During the later years of antiquity,

clay-lined furnaces built of stones replaced the trenches, adequate wind-

pressures produced with bellows ensuring the needed temperatures. Yields

were then somewhat improved."34 Aitchison adds the conjectural thought

(p. 78) that tin smelting was first generally carried on, probably in north-

western Persia, sometime between 1800 and 1600 B.C. The origins of tin

metallurgy, however, are still a mystery.

Properties of tin

The color of tin is white with a bluish tinge, whiter than either silver or

zinc. When solidified from the melt, tin is a mass of crystals with a melting

point of 231.9°C (449.6°F) and a specific gravity of 7.3. Pure tin is soft and

can be cut readily with a knife; it is slightly hardened by hammering.

Chemically, it is moderately active; and though little affected by air at

normal temperatures, it does acquire a superficial tarnish after long ex-

posure to moist conditions. Buried in the earth metallic tin is fairly easily

oxidized to stannic oxide, and this may be the reason why ancient tin

artifacts, if such ever existed, have not survived to tell us more of the early

history of metal.35

33"Notes on the prehistoric metallurgy of copper and bronze . . .p. 17.

3iA history ofmetals, Vol. I, p. 186.35In the late Bronze Age shipwreck found off Cape Gelidonya in southwestern Anatolia (mentioned in foot-

note 29), a number of piles of white powdery tin oxide were found under the copper ingots. The finders believe

that these white deposits were the corrosion residues of tin ingots. This find indicates that, as early as the 13th

century B.C., copper and tin metallurgies were separate and distinct; and it also explains why ingots of early

tin are seldom found.

8

In making bronze, the effect of tin is to make the alloy harder than pure

copper; in higher proportions it embrittles it. This behavior would be of

interest to tool and weapons makers, but to the makers of ceremonial

vessels it would be of only minor concern.

LEAD

Sources andproperties

The third and least of the principal alloy constituents of bronze is in

nature more abundant than tin but less than copper, making up about

0.001 percent of the earth's crust. 36 Like tin, lead has only one important

ore which is galena or lead sulfide, PbS. Although it occurs in many types

of deposits, galena is readily recognized by its lead-gray metallic luster and

its perfect cubic cleavage. The separation of lead from galena, in spite of

the need to rid it of sulfur, is a very simple matter, and as Aitchison has

pointed out, that must be the very reason why so few details of lead tech-

nology survive from antiquity.37 There were no problems. It was only neces-

sary to roast galena in contact with wood or charcoal. In the process, which

is essentially that described by both Agricola38 and Biringuccio39, the sulfur

component of galena is burned off as sulfur dioxide; the lead oxide reacts

with unconverted lead sulfide and carbon, and metallic lead is released.

Thanks to its high specific gravity (11.35) and its liquidity, metallic lead

collects in a pool at the bottom of the hearth.40 This crude lead often con-

tains silver; and, as metallurgy developed, the two were separated by a

process known as cupellation. In fact, the smelting of galena was almost

invariably undertaken for the sole purpose of recovering the silver content,

and the process is described in some detail by E. A. Smith who got his in-

formation from first-hand sources in China.41 Lead metallurgy is probably

much earlier than tin, possibly even earlier than copper; and according to

Wertime it may have started in the Near East as early as the 5th or 4th

millennium B.C.42

The properties of lead are familiar. Its melting point of 327.4°C is higher

than tin but much lower than copper. Its outstanding property, as comparedwith tin and copper, is its high specific gravity which is 11.35. Among the

36 Fleischer, U.S. Geological Survey Circular No. 285, p. 5.37A history ofmetals, Vol. I, pp. 184-185.ssDe Re Metallica, Hoover and Hoover Trans. (1950), pp. 390-394.SSpyrotechnia, Smith and Gnudi Trans. (1943), pp. 54-59.40According to Wertime ("Man's first encounters with metallurgy," p. 1262), however, the smelting of leadores is not always so simple. In the traditional Iranian practice, lead ores are reduced by a process in which ironore (haematite) is added to the charge, and lead and iron were produced simultaneously. There is no evidence,however, that this process was used in early Chinese metallurgy.41"Early Chinese metallurgy," p. 29; see also Gettens and Waring, "The composition of some ancientPersian and other Near Eastern silver objects."42"Man"s first encounters with metallurgy," p. 1262.

9

three constituents of bronze, lead is, in many respects, the most resistant to

corrosion.43

THE BRONZE AGE

The significance of the Bronze Age in the history of man goes far beyondthe simple fact that it was the time when he began to use metal tools andweapons instead of stone. It was in fact the time when civilization began.

Following the domestication of plants and animals that made the Neolithic

the most important single step in all history, people began to settle down in

fertile areas and become farmers; and in the Bronze Age the natural

corollary to this was the gradual formation of cities. Cities in turn intro-

duced a whole new way of living. As William Howells has put it, "There

was a new division of labor, so that some lived as artisans entirely; newkinds of work appeared, and these called for trade in materials; and in the

centers of work and trade, the cities, people lived close together in blocks of

houses. Over such a large population rose rulers of yet unheard-of power,

able to construct temples and palaces. And through all this ran knowledge,

writing, and mathematics."44 Such was the setting in which bronze tech-

nology took its place in human experience.

Long ago, the late Carl Whiting Bishop pointed out that mankind as a

whole did not pass through the Bronze Age, but that the early use of bronze

appears to have been confined to a belt running across the North Temperate

Zone of the Eastern Hemisphere.45 Beginning apparently in southwest

Asia some six or seven thousand years ago, it lasted in some regions, as for

example in China, almost down to the B.C./A.D. milestone. Old Worldarchaeologists seem to agree that the smelting of metals, especially copper,

was first practiced in western Asia in the area known as the Fertile Crescent.

Wertime has recently pointed out that conditions for the emergence of

metallurgy were best fulfilled in the semiarid mineralized zones of the Anato-

lian and Iranian plateaus where fuels as well as ores were abundant.46

According to Cyril S. Smith the use of native copper in this area began

perhaps as early as the ninth millennium B.C.;annealing was known per-

haps as early as the fifth millennium.47 The reduction of oxide ores and the

melting and casting of metals was developed during the fourth millennium

B.C., while the roasting and smelting of sulfide ores did not take place until

about the second millennium.

43Lead objects often survive well, but lead inclusions in cast bronze are sometimes the first components to

corrode.

Mankind so far, p. 230.45Manfrom the farthest past, p. 293.46"Man's first encounter with metallurgy," p. 1258.47"Materials and the development of civilization and science," p. 910.

10

The Bronze Age reached China relatively late, apparently after 2000 B.C.

Whether the vast stretches of desert, steppe and mountain that intervene

delayed the arrival of influences from western Asia, or whether it was

simply a belated but independent cultural development, we still do not know.

Coming generations of archaeologists and sinologists must one day answer

these questions. But when it appeared, independently or otherwise, in the

valley of the Yellow River, the Bronze Age took root and flourished in a

way that still astonishes students of early cultures.

11

K \l %. r. <,

— v

',(K)I rn

iu 1*1

A

O

o

A

m QJ IK fill

65

100J

200Krn

Figure 1

Map of copper and tin mines in northern China centered on An-yang. Copper mines; A Tin mines;

ffl Hunting preserve; 'TfTownsite. After Motonosuke Amano, "Mining and Agriculture in the Yin Dynasty."

Tolw Gakuho, Kyoto, No. 23, March, 1953, p. 236.

12

CHAPTER II

BRONZE IN ANCIENT CHINA

Bearing in mind the foregoing observations with respect to bronze and

its component metals, and the Bronze Age in general, we now turn to the East

to see how they apply to the history of bronze technology in ancient China.

It is a truism that the Chinese had the longest Bronze Age and developed

the highest levels of skill in bronze technology of any people on earth. Theevidence of this is all about us in the museums of the world. Nor does this

statement in any way belittle the accomplishments of other peoples;

Sumerians, Egyptians, Iranians, Greeks, and many others had bronze

cultures of high order; and the monuments they have left behind commandthe respect and admiration of all. But from about 1500 B.C. for a period of

some 12 or 13 centuries the Chinese produced bronzes conceived with an

imagination and executed with a skill that have never been surpassed. Such

a sustained high-level performance in any one field of technology or of the

arts is almost without parallel in the annals of man.

Exactly how and when this all began we do not know. Historians of metal-

lurgy, generally recognizing the origins of bronze technology in the Fertile

Crescent of western Asia, believe that the knowledge of copper and bronze

was taken from there eastward across Asia to China perhaps in the second

millennium B.C. Tending to lend support to this theory is the fact that, so

far as we know, the earliest Chinese bronzes are among the finest; and the

bronze culture of China seems to have come into being in full flower. Notraces of primitive or experimental workmanship have yet been found;48 but

since Chinese archaeology is only in its infancy, we may, as indicated in

the previous chapter, have to wait many years before the full story of

Chinese prehistory is known.49

48The Japanese chemist, Tsurumatsu Dono ("The chemical investigation of the ancient metallic culture in theOrient."), in the 1930's carried out chemical analyses on 65 bronze objects, mostly weapons (spearheads,halberds, etc.), which were found at An-yang, or were purportedly from An-yang, and observed that 31 out of65 samples analyzed contain less than 1 percent of tin or none at all; hence he feels that they should be called

copper instead of bronze. Dono concludes from his evidence that in ancient China a Copper Age did precedethe Bronze Age as it did in Egypt and elsewhere. Barnard, however, points out (Bronze casting . . . , p. 184) that

Dono's analyses were all done on limited and unattested material, and there is still insufficient evidence forclaiming a Copper Age for China.49The present state of knowledge in respect to the emergence of civilization in North China seems to be well

summarized by Kwang-chih Chang in a single paragraph in his recent work, The archaeology of ancient China(p. 136). "Whichever chronological scheme we may choose, the fact is that the known beginning of civilization

in China is approximately a millennium and a half later than the initial phases of Near Eastern civilization.

We can also take note of the fact that many essential elements of Chinese civilization, such as bronze metal-lurgy, writing, the horse chariot, human sacrifice, and so forth, had appeared earlier in Mesopotamia. Here,

13

On the other hand, arguments may be advanced to suggest that the

Chinese did, in fact, originate and develop their own bronze culture quite

independently, and that they started out quite without foreign influences.

Basic to this side of the argument is the fact that the forms of Chinese

bronze vessels and bronze weapons cannot be traced to any outside in-

fluences; they do not seem to copy anything known to have existed in earlier

bronze cultures in the West. They do, however, in some ways reflect the

styles and customs of the prehistoric potters of China who preceded them.

And finally, there is no hint in the Shang and Chou writings of any major

cultural movement from the West onto Chinese soil. H. G. Creel who dis-

cussed at some length the origins of Shang culture sums up the problem

well : . . . "whatever the history of bronze making may have been, the Shangpeople and their ancestors, when they made objects of bronze, stamped

them with something of their own culture." 50

It is not our purpose in this book to try to answer the moot question of

the primacy of foreign versus native influences. At the moment it would be

premature to be dogmatic. If the material we present here provides any

hints or suggestions that may lead a future scholar to the ultimate solution

of this fascinating problem, we will consider our efforts worthwhile. In the

meantime, as we set forth the results of our study, we will not shrink from

pointing out any details that point in one direction or the other.

The great body of early Chinese literature is sparse in writings that relate

to the beginnings of bronze technology. Until the 1920's, controlled

excavations in China had never been undertaken; but archaeological

evidence produced by Chinese scholars of the Academia Sinica at An-

yang and elsewhere before World War II, and archaeological researches

now being conducted under the sponsorship of the People's Republic of

China have already opened up many new windows into the metallurgical

past.

We assume that all metals employed by the early Chinese for making

bronze were found within the confines of what is usually considered as

China proper. Even if the technology of winning metals was imported, there

is no evidence that ores or metals were transported to China from west or

south. China herself had the copper, tin and lead ores necessary for the

production of bronze.

then, is the problem of East-West relationships all over again. And again, the problem must be resolved in twoparts. First, did some of the civilizational elements of the Shang dynasty originate in North China, or did they

come from the Near East? Second, were these, if coming from the West, responsible for the emergence of

civilization in North China? The first question must obviously be answered piecemeal, and for many of the

elements, there is no definite evidence one way or the other. As to the second question, many scholars argue

that civilization came to China as a result of stimulus diffusion from the Near East, principally because they

think it came suddenly and without previous foundation."50 ".

. . bronze in the Shang period," p. 5 1

.

14

Copper in China

Although China is not regarded as an important copper-producing

country, obviously there was sufficient copper ore to supply the demands of

primitive metallurgy. It is commonly said that copper is found in almost

every province; yet no first rank deposits have ever been reported. Accord-

ing to H. K. Tong, the chief copper production centers of modern China are

Yunnan, Szechwan and Sikang provinces.51 The estimated reserves of

copper in China are about a million metric tons but in the last century

China produced only a few hundred tons of refined copper per year, which

is much less than it needs. 52

The Shang capital at An-yang is known to have been an important

bronze casting center; and since some of the vessels in the Freer collection

are presumed to have come from there, the question of the proximity of

An-yang to ore deposits of copper, tin and lead in that region naturally

arises. According to M. Amano there is information from earlier Chinese

writings and gazetteers which tells us that deposits of both tin and copper

existed in northern Honan and in Hopei provinces within a radius of 50 miles

of An-yang.53 Amano shows a map (fig. 1) on which these mines are located

and their distances from An-yang are indicated in kilometers; also Barnard

shows the distribution of tin and copper ores up to the 400 km. radius

and of copper ores to beyond the 3000 km. radius based on a map shown by

Shih Chang-ju.54

Barnard in commenting on the nearer sources (200-300 km.) to An-yang

says, ".. . the comparatively thick distribution of both copper and tin

deposits in the Shensi-Honan area offers a particularly important basis for

argument towards Chinese rather than non-Chinese origins of the art of

bronze casting." 55

Although it is said 56 that native copper occurs in Honan province in

China, and we can suppose that the early Chinese were acquainted with it,

no reports of early objects made of native copper are known to the writer.

There is no reason yet to believe that it was an important source of copper

for making bronze.51China handbook 1937-45, p. 393.52Couling (The Encyclopedia Sinica, p. 368) says that the copper deposits of Yunnan have yielded, in the

more recent past, practically all the copper required for minting and for other purposes in China and Burma.The area about T'ung-ch'uan in northeast Yunnan has been the chief source of supply; mines there wereorganized in 1697 and mining was made a government monopoly in 1738 to supply copper for the increased

demand for brass coinage. Couling adds that the ores mined at T'ung-ch'uan are malachite, azurite, cuprite,

pyrites, and others.5 3"Mining. .

.

," pp. 231-237.54Barnard, Bronze casting . . . , facing p. 50; Shih Chang-ju, "Yin-tai . . .

," facing p. 105; it will probably be a

long time before analytical work on artifacts of early China can be correlated with analysis of ores as it hasbeen done in Germany by Witter and Otto et al. and by Pesta and Pittioni in Austria; see Britton and Richards,"Optical emission spectroscopy . .

.".

55Bronze casting . . . , p. 49.56Carl W. Bishop personal file of notes, section "Copper" Freer Gallery Library.

15

Secondary copper ores

It is likely that most of the copper of ancient China came from the

secondary copper ores already mentioned. Unfortunately, we know of noearly Chinese description of how the ores were recognized. There is cor-

relative evidence from early Chinese wall paintings that the simple carbonate

ores, malachite and azurite, were well known in the Far East. Beginning

with the T'ang dynasty and continuing to the present time, finely powderedbright green malachite and bright blue azurite were employed as paint

pigments.57 Although these colors may have been transported long distances,

they do suggest that secondary copper ores were available in China.

Barnard, basing his researches on former investigations of Orvar Karl-

beck, Shih Chang-ju, and others and on recent Chinese mainland archae-

ological reports, has dealt with early Chinese furnaces, crucibles, and ingot

molds discovered at An-yang and other early sites in China. 58It therefore

seems unnecessary to conjecture further here about ancient smelting

practices.

We may presume that, as in the West, early Chinese metallurgists

produced raw copper in ingot form. We have no evidence, however, of the

existence of ancient copper ingots like those already referred to in the West.

Nor do we know how pure the raw copper was. Probably the molten copper

before pouring into ingot molds was well poled with green wood to remove

excess cupric oxide.

Tin in China

Unlike copper, tin has been fairly abundant in China since ancient times.

The principal tin-producing area is in Yunnan province at a place called

Kochiu in the district of Mengtze, which today is the main center of tin

production in China. Draper says these mines have been worked for some400 years and were probably known in antiquity. 59 Slessor describes primi-

tive mining and metallurgical methods employed in this same area for tin,

lead and zinc (and other metals) that have persisted well into this century.60

According to Mantell and Meng tin has been produced in other provinces,

Kwangsi, Hunan, Fukien, mostly in southern China, and the deposits there

appear to be similar to the stanniferous alluvial workings in the Malay

States.61 C. Y. Hsieh describes the tin placer deposits in this general area

57Wall paintings on which these two pigments were liberally used can be seen at the University Museum,Philadelphia, the Royal Ontario Museum in Toronto, The Museum of Fine Arts, Boston, the William Rock-hill Nelson Gallery, Kansas City, the Metropolitan Museum, New York, and on fragments of wall paintings

from Tun-huang in the Fogg Art Museum, Harvard University.

^Bronze casting . .. , p. 48 ff.

59"The tin industry of Yunnan," pp. 483-491

.

60"Chinese non-ferrous metals."61 Mantell, Tin . . . , p. 79; Meng, "Tin deposits . .

.".

16

(called Fu-ho-chung-chiang).62 There is little knowledge of the origin of

the tin used in the bronze casting centers of the second and first millennia

B.C. There seem to have been no noteworthy deposits of tin in north China,

although Amano indicates on his map that there were tin as well as copper

ores within a radius of a few hundred kilometers of An-yang.63

In spite of the wide use of tin for bronze-making in Shang and Choutimes, hardly any artifacts of pure tin, if they ever existed, have survived.

There are a number of early Chinese objects in collections labeled "pewter,"

but when analyzed they usually turn out to be nearly pure lead. This is

true of that collection of so-called "pewter" objects that were illustrated by

Bishop White in his last work (vide infra). Li Chi, however, illustrates two

tin ingots found at Hsiao-fun near An-yang; and he expresses the belief

that the ingot form is an indication that they were imported from somedistance and brought to the copper smelter for making bronze. 64 Diligent

search should be made in Western collections of ancient Chinese artifacts for

tin ingots or tin objects.

Lead in China

Lead, the third of the principal alloy constituents of ancient Chinese

bronze, apparently was in good supply even in early times. Torgasheff says

that lead ore deposits are found in China in a number of provinces;65 and in

southern and western Yunnan there are several ancient workings and ex-

tensive bodies of old slags. There does not, however, appear to be muchinformation about lead ore deposits in north China in antiquity.

We can presume that, as elsewhere in the world, the principal lead ore

was galena. E. A. Smith says that the smelting of galena was almost in-

variably undertaken for the sole purpose of recovering the silver content,

and he described in some detail the cupellation of crude argentiferous lead

cake to separate the silver.66 It would be interesting to know something

about the occurrence and metallurgy of silver in ancient China because wemight get some clues from it about copper and lead metallurgy.

In early China lead was a casting medium in its own right. Several col-

lections have examples of lead ceremonial vessels in the classic shapes of

Shang and Chou bronzes. Bishop White tells of a group of early Western

Chou "pewter" vessels that were formerly owned by Yamanaka andCompany of New York, and he pictures ten of them in his catalogue.67 One62"Tin placer deposits . .

.".

63Amano, "Mining . . .," p. 236; see also Li Chi, The beginnings . .

. , p. 37.640p. cit., pp. 36-37 and Plate X.G5The mineral industry . . . ,p. 164.66"Early Chinese metallurgy," pp. 29-30.67Bronze culture . .

. , p. 176 and Plates XCVII to C.

17

Figure 2

Figure 3

Figures 2 and 3. Six of a set of ten early Western Chou lead ceremonial vessels, formerly owned by Yamanakaand Company of New York, are now in the Hermitage Foundation Museum in Norfolk, Virginia. These andsimilar vessels in other collections testify to the fairly wide use of metallic lead for casting purposes around

1000 B.C. in China. The two pieces pictured here, yu 44.G.38.E {fig. 2) and kuei 44.G.38.A {fig. 3), have promi-

nent mold marks which indicate they were cast in piece molds in much the same way vessels of bronze were

cast.

18

of these, a libation cup of the type chiieh (SC546) has come into the study

collection of the Freer Gallery (see Chapter IV, figs. 35-38). A spectro-

graph^ analysis shows that this vessel is nearly pure lead. Two pieces of

this same set were purchased by Mr. and Mrs. Philips de Jongh of Eind-

hoven, Holland, and are described by J. Fontein. Chemical analysis of

those pieces showed that the metal is 95.4 percent lead and 4.6 percent

antimony.68 Six pieces of this same set of pewter vessels are now in the

Hermitage Foundation Museum in Norfolk, Virginia (figs. 2 and 3). Thus, the

whereabouts of nine of the ten pewter vessels pictured by Bishop White are

known, but the tenth, a chiieh (No. 8 in Bishop White's catalogue), is still to

be found. More recently the discovery of eight lead ceremonial vessels in

Western Chou tombs at Loyang was reported and pictured by Jo Shin.69

These include one vessel each of the types ku, tsun, chih, yu, ting, li and

chiieh. These leaden vessels are comparable to the group of "pewter"

vessels described by Bishop White.

It is almost certain that, as early as Shang times, lead metallurgy was

distinct from both copper and tin metallurgy. The low silver content of

ancient bronzes and the existence of lead objects testify to this.

Bronze

In spite of the sporadic occurrence of weapons made of nearly pure

copper, the Chinese evidently learned the superior qualities of copper-tin

alloy early in their cultural history, and bronze became predominant in the

second millennium B.C.

Many ancient Chinese bronzes bear an excess of lead. In a number of

primitive style bronze vessels from pre-An-yang sites now in the Royal

Ontario Museum described by Barbara Stephen there is more lead than tin;

in two of them lead is reported in excess of 20 percent.70 The commonChinese knife and spade money of the late Chou dynasty is often loaded

with lead. Both Chikashige and Wang Chin report the analyses of several

early Chinese coins in which lead content is in excess of 40 percent.71 It

seems clear that lead was used to cheapen or adulterate bronze at a very

early date.72

How lead was introduced into the alloy we are still unable to explain.

The wide range of lead in the Freer bronze vessels, where it varies in con-

tent from less than 0.5 percent to over 20 percent (see Chart I, fig. 18), is

68Fontein, "Loden vaatwerk . . .," p. 145.

69"Two Western Chou tombs at Lo-yang."70Stephen, "Early Chinese bronzes . . .

," p. 67.7 Chikashige, Oriental alchemy, p. 72; Wang Chin, "The composition of Wu-shu coinage . .

.".

72According to Tylecote (Metallurgy in archaeology, p. 43) parallels may be found in the development ofbronze metallurgy of the late Bronze Age in England.

19

quite astonishing. The presence of lead in appreciable quantities did serve

a purpose other than saving more costly copper and tin; it helped to lower

the melting point, helped to produce a clean casting, and made for easier

finishing operations like removal of ridges and flaws, and final burnishing.

Zinc in China

In the discussion of the chemical composition of the alloy of the Freer

ceremonial vessels in the next chapter it will be pointed out that those

bronzes confidently attributed to the Shang and Chou dynasties contain

scarcely any zinc but, on the other hand, the bronzes of later periods

often bear zinc in varying amounts. There is reason to believe from our

own analyses that beginning in the Han dynasty and continuing into later

periods the amount of zinc in Chinese bronzes gradually increased. This is

supported by the writings of H. T. Chang who in several articles has dealt

with the early history of zinc in China. Chang presents a table of chemical

analyses done by Wang-chih Tien of five bronze coins cast during the reign

of Wang Mang (A.D. 9-22) which show in addition to tin and lead the

presence of zinc ranging from 0.84 to 4.1 1 percent.73 This indicates that zinc

began to appear in Chinese coinage of Han times about the same time it

began to appear in Roman coinage.74 Slessor makes an unsupported state-

ment that "The process of obtaining zinc metal from blende [sphalerite,

ZnS] by distillation has been known to the Chinese since at least 500 A.D.,

and probably 200 years earlier, as the Sung dynasty coins show about 18%zinc contents. This distillation process was not known in Europe until 1700

and probably came from China originally. Spelter [zinc metal] is used in

China for brass making, and also very largely for debasing the moderncopper coinage.

1,75 The late appearance of zinc in copper alloys of China (as

well as in the West) may be explained in various ways : first, the early metal-

lurgists must have failed to recognize zinc ores like smithsonite, ZnCC>3,

sphalerite, ZnS, and calamine, Zn 4Si207(OH)2-2H20, as a source of metal;

second, in earliest times only easily available ores of copper like malachite

were employed, but as these ores were exhausted miners had to go deeper

in the earth where they encountered primary ores which may have contained

small percentages of zinc; third, zinc may have been introduced by the

necessity to dig deeper for lead ores (galena) which often occur in nature

associated with zinc. The finding of little or no zinc among early Chinese

bronzes in the Freer is in essential agreement with results of spectrometric

analyses made some years ago by the staff of the Research Laboratory for

73Chang, "New research . ..," p. 127; see also Chang, "The beginning of the use of zinc in China."

74See Caley, Orichalcum . .

.

75"Chinese non-ferrous metals," p. 111.

20

Archaeology and the History of Art, Oxford, on about 400 Chinese bronzes

of the Shang, Chou and Han dynasties. Sir Harry Garner who has summar-ized the Oxford analyses ("The composition of Chinese bronzes

,,

) points out

that only five of the bronze vessels analyzed had measurable amounts of

zinc and the highest figure for that element among vessels of the group was2.4 percent (see Appendix II). Sir Harry concludes that the presence of

small amounts of zinc in Shang-Yin bronzes is accidental. Much analytical

work remains to be carried out on dated Han and post-Han bronzes before

the story of zinc in China can accurately be told. One of the great difficulties

in this research will be the finding of dated pieces to analyze.

There is no denying, however, that whatever method of smelting and

alloying was used, ancient Chinese metallurgists were able to produce an

alloy eminently suited for the production of vessels which are today im-

portant documents of a culture and testimony to the skill of their long

forgotten makers.

Western views on early bronze casting techniques in China

References to metals technology in early Chinese literature are so rare as

to be almost nonexistent. There are no wall paintings or stone reliefs com-parable to those which Neuberger says can be seen at Karnak in Egypt,

depicting the casting of a bronze door for the temple, 76 or of scenes in a

casting workshop like that shown on a red figure Greek vase in a Berlin

museum.77 Nearly all of our information comes from the objects themselves

or from ancient foundry and workshop sites in China. Modern foundry

practices in China and Japan may still reflect ancient metals technology,

but we do not know much about what is going on now.

Western scholars, however, have been intrigued by the arts and crafts of

the Far East; and there has been much speculation about how Chinese

bronzes were cast. Until comparatively recent times, collectors of Chinese

bronzes assumed that the ceremonial vessels were cast by cire perdue (lost

wax) process, the method that had been employed in the West going back

to early historic times. This view was affirmed by no less an authority than

the late Stephen W. Bushell, who, writing at the beginning of this century,

states categorically, "Chinese bronzes have always, as far back as we have

any record, been executed by the cire perdue process, and finished, whennecessary, with the hammer, burin, and chisel." 78 As late as 1931, W. F.

Collins expressed the same sentiment, ".. . it is believed by many that all

early Chinese bronzes were cast by the cire perdue process, considered by76 777e technical arts . . . , pp. 54, 55.11Op. cit.,p. 58.78 Chinese art, Vol. I, p. 61

.

21

some to have been invented by the Egyptians. 1179 These widely accepted

views were supported by the fact that no exact duplicate castings have

ever been found except among bronze mirrors which could be made in

simple bivalve molds.

Apparently the first critical review of the problem was made by the late

Professor W. Perceval Yetts, in his chapter on "Techniques of Bronze

Casting" in the Eumorfopoulos Catalogue. In introducing the subject,

Yetts laments that he spent much time in fruitless search of the Chinese

literature in order to unlock the secrets of bronze technology. He says, "Nocontemporary account of technical processes survives, nor is it likely that

one ever existed.1180

Yetts first speculates about the proportions of copper to tin required in the

recipes for six different classes of bronze objects that are set forth in the

often quoted passage in the K'ao-kung-chi of the Chou-li, which purports to

be of Chou dynasty origin. In the Six Recipes the proportions of copper andtin are as follows: for cauldrons and bells, 5 to 1

;

81 for axes, 4 to 1 ; for hal-

berds and spears, 3 to 1 ; for swords and knives, 2 to 1 ; for erasing knives

and arrows, 3 to 2; for mirrors and specula, 1 to 1. Yetts questions the Choudynasty origin of the Chou-li and dismisses the Six Recipes with the state-

ment: "The question is whether these figures given in the Chou-li should be

regarded as traditional data, based on actual practice, or as theoretical

standards formulated by the author. 1182 and further he concludes: "In short,

the likelihood is that the alloys used in feudal China conformed to no uni-

form standard, but varied according to the accidental chances of supply and

the rough empirical knowledge of craftsmen. 1183 Only a small number of

chemical analyses of ancient bronzes were available to Yetts, but in the

light of the widely varying composition of bronze vessels in the Freer

collection Yetts1

words are truly prophetic.84

Although Yetts like his contemporaries believed that Chinese bronze

ceremonial vessels were cast by the cire perdue process, he had certain

79"The Corrosion of . ..

," p. 25.80Yetts, Eumorfopoulos, Vol. I, p. 35.81The ratio of copper to tin in the large Chou dynasty bronze bell (chung) in the Freer collection (41.9) is

approximately 5:1. Analysis shows Cu 82.1%; Sn 16.3%; total 98.4%.s '

2Op. cit., p. 34.S3Op. cit., p. 35.84Three of the Japanese chemists who have collectively done most of the analytical work on ancient Chinesebronzes have compared their own findings with the Six Recipes of the Chou-li K'ao-kung-chi, and have ex-

pressed divergent conclusions. Chikashige felt that broadly interpreted, "all the 'Six Receipts1

become generally

speaking reasonable and correct." (Oriental alchemy, p. 62). Dono, however, who analyzed many early knives

and halberds found little relationship between their composition and the proportions dictated for suchweapons in the Six Recipes ("The chemical investigation of the ancient metallic culture in the Orient,"

p. 301 ff.). Tanabe also mentions the "Six Receipts;" but apparently, he finds no parallels between his com-position data and the Six Recipes ("A study on the chemical compositions . . . ", p. 267). Barnard, who hascritically reviewed the "Six Formulae," dismisses them with the statement that, "they are but a numericaljuggling of facts little understood by their author." (Bronze casting . . . , p. 12).

22

reservations about it. He had observed the "seams" on ancient bronzes,

which are now recognized as mold joins; and he offered a modification of the

method of mold preparation to explain them. But Yetts1

proposals on model

preparation are highly conjectural and over-simplified; and they take no

account of preparation of molds for under-cut details of form or design;

or of the casting of vessels with high-placed bottoms; or of the attachment

of handles and swing bails. Recent observations on bronzes in Western

collections and field archaeological discoveries in China have raised further

serious doubts about the employment of the cire perdue process of bronze

casting in early China.

Mold fragments

An entirely new approach to bronze casting in China was introduced by

the discovery in the late twenties of this century at Hsiao-t'un and other

places in the neighborhood of An-yang of fragments of baked clay with the

negative impressions of typical bronze ceremonial vessel patterns on their

inside concave surfaces. These fragments are obviously broken bits of

piece molds for casting bronze. Moreover, many of them have flat edges

prepared with mortises and tenons or keys which show they were actually

parts of piece molds, and not just fragments from a broken eire-perdue waste

mold. An-yang mold fragments seem first to be mentioned by Liu Yu-hsia

in Preliminary reports of the excavations at An-yang (Part I V) published in

1933, and some of them which came from Hsiao-t'un near An-yang are

pictured by Li Chi in his more recent summary account of archaeological in-

vestigations of Shang dynasty sites in the An-yang district since 1928. 85

Twenty-six mold fragments from Yeh, the An-yang area, are well illustrated

by Huang Chun in a catalogue of An-yang antiquities offered for sale in

Peipingin 1937.86

In 1934 Orvar Karlbeck, a Swedish engineer visiting An-yang, learned of

these mold fragments and subsequently acquired from small local dealers a

total of 170 of these, which he later described (1935) and pictured in

some detail. 87 Karlbeck never learned the exact origin of these fragments,

but he believes they were dug up near a village west of Hsiao-t'un by farmers

who had chanced upon the site of an ancient bronze foundry

.

88 The author had

the rare privilege of examining Karlbeck's mold fragments in Stockholm in

1959 {fig. 4). Karlbeck believes that these fragments represent at least 85

different molds and that the molds were used for the direct casting of bronze85Liu Yu-hsia, "Study of bronze casting . . . .," pp. 681-696; Li Chi, The beginnings . . . , Plates XIII and XIV.86Huang Chiin, Yeh-chung . .

. , p. 43 ff.

87"An-yang moulds."88Karlbeck says (private communication) that he purchased a few fragments from a small boy directly in

Hsiao-t'un itself.

23

Figure 4

Fragments of clay mold sections from An-yang in the Collection of

the Museum of Far Eastern Antiquities, Stockholm. (Karlbeck,

"An-yang Molds," BMFEA No. 7, pp. 39-60 [1935]. Courtesy,

Museum of Far Eastern Antiquities, Stockholm .(About § natural

size.)

Figure 5

Fragments of clay mold sections in the Department of Oriental

Antiquities, British Museum. The one on the right has a join surface

which bears the negative form of a flange often seen on bronze

vessels. (Courtesy, British Museum, Department of Oriental Antiqi-

ties. About § natural size.)

24

vessels, and he shows diagrams of piece mold assemblies to support his

thesis. He believes further that the ornamental details were not produced

by casting from a model but were incised directly in the clay\ He does not

believe that wax played any part in model or mold preparation. Karlbeck

bases his belief in direct casting on four principal observations: (1) The

inside of the mold fragments are gray which is characteristic of reddish

baked clay that has been in direct contact with molten metal. (2) On one

fragment a bit of bronze metal still adheres to a narrow groove in the

design. (3) Many fragments show traces and patches of soot only on the

inside surface which Karlbeck believes is indication that soot was added

intentionally to permit the mold section to release easily from the metal

after casting. (4) The mold sections are porous, which is a necessary charac-

teristic when metal is to be poured directly into the mold, but not necessary

if the section is to be used in the preparation of a wax model.

In addition to the An-yang mold fragments in Stockholm there are also

numerous fragments from the An-yang region in the collections of the

Academia Sinica at Nankang in Taiwan. Pictures of many of these have

been published.

Other collections of the same kind of mold fragments, probably from the

An-yang area, exist in the West. There are a handful in the Department of

Oriental Antiquities at the British Museum {fig. 5). The Museum of Fine

Arts, Boston, has fragments of a piece mold for a vessel of the type ku; and

the Fogg Art Museum of Harvard University possesses two mold fragments,

one of a ku and the other probably of a kuei; although well weathered, both

fragments show a portion of a mortised edge. The Seattle Art Museum has

a mold fragment of a chiieh.

In the Royal Ontario Museum in Toronto is an important collection of

An-yang mold fragments bought in China by Bishop White. The late

William Todd, Chief Conservator of that Museum, said there are a total of

134 fragments in the collection of which 33 show keys. The Royal Ontario

Museum has generously made these fragments available for examination at

the Freer Laboratory where the following characteristics have been noted:

1. The mold material is baked earthenware, like soft brick.

2. Some fragments have two distinct layers of clay, an inner fine grained

layer, and an outer coarser one (fig. 6).

3. The grain structure of the mold fragments, even of the inner layer, is

uneven and gritty and does not appear to be made from fine, well-

levigated clay.

4. On some fragments a portion of the intaglio design is crisp and non-

weathered, but in closely adjacent areas, the mold surface is damaged as

25

Figure 7

Figure 9 Figure 10

Figures 6-10. Baked earth mold sections from a collection of about 130 fragments in the Royal Ontario

Museum, Toronto, Canada. Figure 6. A mold section (No. 28; about half size) from the waist of a ku. Thesquare key or tenon on the top and the triangular mortises on the sides serve to keep the sections in register.

Two distinct layers of earthen material make up the section, an inner layer of fine earth and an outer one of

coarser earth. Figure 7. A mold fragment (No. 41 ; about half size) with a positive impression cast in plaster.

Parts of the mold parted cleanly from the bronze but other areas broke away during the separation. This is

evidence that the mold sections were not intended for reuse. Figure 8. Detail of another fragment (x 1.5)

showing the general character of the negative design, especially the depth and straight sides of the channels that

form the relief decor elements in the finished bronze. Figure 9. A mold section (half size) showing a portion of

a negative mold for a tao-tieh mask. Figure 10. Detail of an earthen mold section (No. 949.143.3; x 1.5)

showing the character of the decor lines. The edges of the channel lines are slightly depressed like wheel tracks

in a dirt road. (Courtesy, Royal Ontario Museum, University of Toronto.)

26

if it had not pulled away cleanly from the metal {fig. 7). This could in-

dicate that the mold piece was not intended to be used again, but is in-

deed part of a waste mold.

5. Some of the grooves of the design are deeply and narrowly modeled or

cut {figs. 8 and 9). This must reflect the nature and character of the

material in which the original modeling was done.

6. On two of the fragments (ROM 949.143.3 and ROM 56) the grooves

of the wide lines of the pattern are cut deeper at the edges than in the

center {fig. 10).

Close study of these Royal Ontario Museum mold fragments tends to

support the direct casting theory advanced by Karlbeck. It appears that

the modeling of the decor was done negatively and directly in the clay. It

is difficult to believe, because of the narrowness and depth of the grooves,

and also from the coarseness of the baked clay, that the negative mold was

parted from a wax positive.

There are others, however, who believe that the piece mold assembly was

made and used in a very different way. Herbert Maryon and H. J. Plender-

leith present a reconstruction of a piece mold assembly for casting a bronze

tsim, and give their version of the casting process.89 They believe that the

original modeling was done in wax applied to a core and that the piece mold

was built around the wax models. Eventually the wax was melted out as in

the classical cire-perdue process.

The most detailed and comprehensive study of this casting problem is

the work of Noel Barnard, Bronze casting and bronze alloys in ancient

China. Barnard accepts Karlbeck's theory of piece mold casting and carries

the thesis much further. He believes that Chinese metallurgy was not deriva-

tive but was an indigenous and independent development. Based on mold

fragments from China, much like those described above, and with

reference to actual models in clay excavated from An-yang, Barnard

develops a process of mold making which takes into account also the in-

vestigations of Shih Chang-ju, a member of Academia Sinica who partici-

pated in the early excavations at An-yang. 90 Apparently he was the first

Chinese to realize the importance of the mold fragments. The first step in

preparing the mold is the construction of a model in clay of the ultimate

article.91 All decor elements were pressed or cut into the model. The model

was then passed to the artisan who specialized in the building of sectional

molds. Barnard, with aid of drawings, develops an elaborate theory of moldconstruction employing multiple sections with horizontal as well as vertical

89"Fine metal-work," p. 628 ff.

90Barnard, Bronze casting . .. ,Chapter V, p. 1 10 ff. ; Shih Chang-ju, "Yin-tai . .

.".

91 Barnard, Bronze casting . . . , p. 65.

27

joins. He believes that at An-yang and even in later times the principle of

"sectionalism11

, as he terms it, involving the production of decor impressions

in blocks of clay from master patterns, played a major role in model andmold preparation. In his Chapter V dealing with "Sectional moulds and the

significance of sectionalism," Barnard uses one of the best known vessels in

our collection, thefang-i (30.54), and vessels in other collections as examples.

With the use of interpretive line drawings he attempts to explain how the

molds for them were made.92

The complexity of Barnard's mold reconstructions had been criticized by

persons knowledgeable in foundry practice93 who believe that metal foun-

ders in ancient times, as they do today, made molds in the simplest waypossible. Interesting confirmation, however, of some of Barnard's ideas on"sectionalism" has since appeared in reports on the Hou-ma foundry site of

Chan-kuo date in Shansi province, notably the one by Chang Han where

master patterns and mold sections are pictured (vide infra).

Barnard goes further. He expresses strongly his conviction that the cire

perdue method of casting bronze was unknown in China in the Shang and

Chou periods, and he believes it did not come into China until the advent of

Buddhism.

Wilma Fairbank strongly supports the piece mold theory but points out

that the mastery of bronze technology by the Shang people was an out-

growth of their earlier mastery of ceramic technology.94 She says that dis-

coveries at An-yang suggest there was a direct transition from making

pottery vessels to producing the same or similar shapes in bronze; and by

the use of piece molds, this would have been feasible. Mrs. Fairbank finally

points out that, in the fully developed Shang vessel decor, the animal forms

are broad plastic elements protruding from the vessel surface, and they

appear to be formed by building up clay on the positive (the model), not by

carving in the negative (the molds).

The most recent study on An-yang piece molds is a monograph with the

English subtitle, Studies of the bronze Ku-beaker edited by Li Chi, Shih

Chang-ju and Kao Chu-hsun and issued in 1964 from Nankang, Taiwan.

Following a general description by Wan Chia-Pao of 42 specimens of muchcorroded ku excavated at Hsiao-t'un and Hou-chia-chuang, and of 26

sections of earthen molds used for casting ku, he concluded : "The successive

steps of the casting process appear to have been :

92Barnard admits later (private communication) that his reconstructions err in one main point - the masterpatterns were engraved with positive decoration, not negative, hence the repeated designs pulled from themwere inserted directly into the molds and not into the models.93Herbert Maryon, in his review of Barnard, Bronze casting . . .; also Cyril S. Smith in his review of the samebook.94'Tiece-mold craftsmanship . .

.".

28

1. A model was first made. The pattern was incised partly on the model

and partly afterward on the mould according to the nature of the pattern.

The model was then baked.

2. To form the mould, a fiat piece of clay with a thickness of about

15 mm. was pressed against the model. The edges of the mould were

finished with a knife, and some kind of release agent was applied to

prevent the adhesion with the adjacent mould. The divisions of the mould,

and the number of moulds were all determined by the shape of the pattern.

The mould was then baked.

3. The moulds were assembled around the core and the assembly was

bound with grass rope and/or put into a sand box in order to assure

rigidity. The moulds were assembled upside down as evidenced by the

use of chaplets to support the suspended ring foot core.

4. The core was made by scraping down the surface of the model. Thethickness of clay scraped away was equal to, or a little thicker than, the

wall of the bronze to be cast.

5. Finally, the molten bronze was poured into the moulds. There is no

evidence to show to what temperature the moulds were heated when the

molten bronze was poured. But it must have been a very high temperature

to avoid sudden cooling of the molten bronze on contact with the mould.

The clear pattern of the Shang bronzes reveals that no such loss of

fluidity took place during this critical step of the casting process." 95

So far the little evidence about Shang casting methods that can be gleaned

from study of the few and scattered sections of piece molds that have comefrom the restricted area of An-yang has been subjected to varying and even

contradictory interpretations by the several investigators, mainly Western

scholars, who have studied them. They have been at great disadvantage

because of the meagerness of the evidence.

A new chapter in the study of ancient methods of bronze casting was

opened by the discovery and excavation in recent years by Chinese archae-

ologists of an Eastern Chou bronze foundry site near Hou-ma in Shansi

province. As reported by Chang Shou-chung in the Chinese archaeological

journal Wen Wu, literally thousands of pottery fragments, clay mold frag-

ments, bronze weapons, bone and shell artifacts, and coins were uncovered

during the 1956-59 excavation periods.96 In a later article Chang Hanreported that excavations west of Hou-ma in 1956 had uncovered 50 to 60

thousand fragments of clay molds dating from the Ch'un-Ch'iu through the

middle of the Chan-Kuo periods.97 Included in the finds are separate molds95From the English Summary by Wan Chia-Pao, Part I, "The process of casting the Ku-beaker," pp. 121-122.96Chang Shou-chung, "1959 . . .

."

97Chang Han, "Hou-ma "

29

for vessel bodies, vessel covers, and handles and legs of ting. There are also

illustrations in the report of piece molds for small human figures in which

triangular mortises and tenons to hold mold sections in position plainly show.

In a third illustrated report on the Hou-ma site in Wen Wu by Chang Wan-chung, more detailed information on piece mold construction is given. 98 In a

certain location in the southern portion ofthe Hou-ma excavations many moldfragments were found in excellent condition and some fitted together quite

perfectly. A few of the mold fragments have been closely examined andanalyzed to determine the sand-clay ratio. Chang differentiates between

"mother" molds or models, and "inner" and "outer" molds. There appear

to be two kinds of "mother" molds; first, those which are seamless and

appear to be made by hand; and secondly, models with a seam running

through the design which indicates that they were made in a mold. Theouter mold was reproduced from the "mother" mold and was used directly

in making castings. Chang Wan-chung says that when the "outer" mold was

taken from the solid "mother" mold and before it was thoroughly dried, it

was cut into several sections, depending upon the shape of the object. Thepurpose of the several pieces was to facilitate removal of the mold from the

bronze after casting. The outer mold was generally used only once. Theinner molds, or cores, were made of coarser clay mixed with vegetable fiber

so that during firing it became porous and permitted escape of air during

casting. On many of the mold sections that are pictured one can see a

number of small triangular mortises and tenons which served presumably to

keep the mold sections in proper register. It is also clear that some of the

sections are provided with wedge-shaped pouring cups. Although it is only

possible here to summarize briefly Chang Wan-chung's long report, it

should be mentioned that he uses the archaeological evidence to reconstruct

the method of casting a bronze ting. He says that the handles and legs of a

ting were separately cast each with a clay core and that the members were

then inserted in suitable openings in the piece mold for the vessel proper

in such a way that when metal was later poured to form the vessel it locked

to the members to complete the assembly. That some such method ofjoining

members of a bronze was actually used is amply supported by observations

that will later be reported (Chapter IV) in descriptions of construction of

our ceremonial vessels.

These recent discoveries by field archaeologists of the People's Republic

of China have come to the attention of Western scholars only through

brief and entirely inadequate and poorly illustrated preliminary reports.

They are more tantalizing than informative, but they do raise hopes that,

98Chang Wan-chung, "Hou-ma . . .

30

perhaps in another generation, the mysteries of bronze casting in early

China will be unveiled.

There is good reason to believe that direct casting of bronze in earthen

piece molds, following in many respects the tradition of ancient Chinese

foundrymen, is still carried on in Japan. Noel Barnard tells of visiting a

foundry in Tokyo where small statuary, Buddhist figures and even copies of

Chinese ritual vessels, are cast in piece molds." As Barnard describes it the

molds are built around models constructed of wood, clay, plaster or any

material in which the artist chooses to work. After the mold sections are

removed from the model, the inside of each section is filled with a layer of

sand-clay-fiber mixture to the intended thickness of the casting space. Aporous sand mixture is then piled into the cavity of each mold section and

when the sections are assembled this filling becomes the core. When the

assembly is sufficiently dried the outer molds are dismantled, the fibrous

sheeting which has formed the casting space is stripped off and discarded;

chaplets or spacers are inserted at suitable intervals, and the mold sections

are again assembled, and at the same time provided with ingates and air

vents. After the mold assembly is baked it is ready for pouring the metal.

These piece molds are used only for making a single casting because they

break up and are damaged on removal from the cooled metal. Wax is not

employed in any step in the process.

The writer, at the Fourth Sculpture Casting Conference (Spring, 1966) at

the University of Kansas (Lawrence), witnessed a demonstration by Daiyu

Nishi of Tokyo University of Arts of a primitive process of casting em-

ploying earthen piece molds which seemed very similar to the casting pro-

cess described by Barnard.

The preceding paragraphs attempt to cover briefly, though perhaps in-

adequately, several facets of early metals technology and the views of someof the scholars who have given thought to methods of bronze vessel fabrica-

tion. In the chapters that follow an attempt will be made to summarize

composition and construction characteristics of 122 ceremonial vessels in

the Freer collection, which over the past five years have been subjected to

closer inspection and to more technical scrutiny than any comparable group

of objects in the world. The findings will test all of the theories on ancient

bronze casting that have been put forward. They are sure to raise morequestions than they answer, but that is as it should be.

"Barnard, "Book review: Cheng, Prehistoric China . ..," pp. 249-252.

31

CHAPTER III

COMPOSITION

In beginning the technical investigation of our bronze ceremonial

vessels, it was recognized that the question of alloy composition is a funda-

mental matter mainly because the facts of composition are essential to an

exact description of each object. Alloy composition had to be considered in

connection with numerous questions that undoubtedly would arise later in

the investigation such as those that relate to sources of raw materials,

physical properties of the alloy, methods of fabrication, corrosion resistance

and corrosion products, and others. It was felt necessary to get an affirma-

tive or negative answer to the frequently asked question: Is there any rela-

tionship between composition and age and provenience? It was further

realized that since many of these bronze vessels are judged on the basis of

style and inscriptions to be genuine documents of early Chinese culture,

knowledge of metal composition would be an important contribution to

the history of the beginnings of metal technology, not only in China but

also in the wide world. These vessels have been selected to serve as standards

against which objects of similar kind are to be compared, not only in

style and decor, but also in substance.

The study of the composition of early metal objects of China has been

carried on for some years by many investigators; in fact, there are in exist-

ence many more published analyses of Chinese bronzes than is commonlyrealized, even by Sinologists. A recent survey made from the Freer Labora-

tory has uncovered some forty sources in articles and unpublished theses, in

which over four hundred quantitative chemical analyses are reported. Thebulk of them relate to bronze weapons of the Shang and Chou dynasties.

The next largest number of analyses are of bronze mirrors of the late Chouand of the Han period. There are also scattered analyses of tools, musical

instruments, vehicle and harness fittings, culinary utensils and of coins.100

Between thirty and forty analyses relate to ceremonial vessels, but many of

them were done on small fragments, which are difficult to date and to place.

In only a few instances is reliable information about provenience given,

and only about half of the ceremonial vessels analyzed in these reports are

illustrated. 101 With but a few exceptions, all of the analyses on Chinese100Many of these scattered analyses are tabulated by Barnard, Bronze casting . . . , Chapter 6.101Analyses of a small number of ceremonial vessels which are reported in the literature with photograph arelisted in Appendix II.

33

bronzes found in published reports were done by the classical methods of

wet analysis, hence only major and minor constituents are reported. Onlyrecently have spectrometric analyses for trace elements begun to be reported.

During the course of our investigations, it was learned that a similar

study had been undertaken recently in Japan by Dr. Giichi Tanabe on a

large number of early bronzes found in Japan.102 These artifacts, mostly

mirrors, bells (dotaku), weapons, and implements date from the Yayoi andTumulus periods in Japan, and some were from China and Mongolia.

Samples weighing 0.1-0.5 grams were obtained by drilling into the un-

corroded or less corroded center of the object. Each analysis reported by

Tanabe was done in two steps: the main constituents, copper, tin, and lead

were determined by the classical methods of wet analysis; the minor and

trace elements were estimated by semiquantitative emission spectrometry.

The analytical results are all presented in tabular form. Tanabe's report is

especially interesting, useful, and timely because the scope of the problem,

the types of objects analyzed and the analytical methods used are similar to

the parallel investigation on early Chinese bronzes that has been carried on

at the Freer. Tanabe's report will be referred to frequently in the following

pages.

In recent years there has been an upsurge of interest in the chemical

analysis of antiquities of all kinds. In addition to the classical methods of

analysis which require the taking of small samples from the object being

analyzed, there are being advocated and employed a variety of instrumental

methods of analysis using physical principles. Some of these may be used

directly on the object without taking a sample; others require a sample, but

the sample is not destroyed in the analysis. Among these so-called non-

destructive methods are X-ray fluorescence analysis, neutron activation

analysis and analysis using the electron beam microprobe technique.

These physical or indirect methods are useful under certain conditions, but

they require special sample preparation techniques and skilled personnel

well acquainted with archaeological problems. In spite of limitations in the

use of such modern instrumental methods of analysis, they are bound even-

tually to influence the course of "archaeological chemistry." Advance of

knowledge into the field of ancient metal technology in China will be limited

only by the availability of objects and specimens for technical study.

The analyticalprocedure

In the study undertaken here it was decided for practical reasons to carry

out the analyses in two steps : wet chemical quantitative analysis for major102"A study on the chemical compositions . .

.".

34

constituents, and semiquantitative analysis by optical emission spectro-

metry (spectrography) for minor and trace constituents. For purposes of

these analyses a major constituent is one in amount above 1.0 percent; a

minor constituent in the range of 1 .0 to 0.01 percent, and a trace constituent

is in amount below 0.01 percent. The accuracy demanded for major con-

stituents, especially copper, tin and lead, required their separation and in-

dividual determination to within 1 percent of their true amounts. Limita-

tions in regard to the size of the samples that could be taken required the

use of spectrometric analysis for the estimation of minor and trace elements.

Because of its high sensitivity, the spectrometric method is not suitable for

major constituents, which is another reason they had to be estimated by

wet methods. Moreover, the high content of tin and lead, and the wide range

in the relative proportions of these elements, made the preparation of suit-

able spectrometric standards difficult. Both wet and spectrometric analyses

made on the Freer bronze vessels are reported in Tables 1A and 1 B.

Sampling

Samples of interior metal were taken by drilling into the bronze with

small high speed twist drills ranging in size from Numbers 60 to 44 (0.10

to 0.21 cm. diameter), usually rotated with an electric motor with

flexible shaft (figs. 11 and 12) but sometimes with a small hand drill. Mostof the bronzes sampled were corroded on the surface so that care was taken

to get the sample from deep in the pure metal. The samples were usually

taken from an inconspicuous place like the rim of the foot, under a flange, or

the underside of a leg (fig. 13). The difficulty of taking a single sample that

would represent the all-over composition, that is a "fair sample," was fully

realized.103 In some cases samples were taken from two or more places on a

single object to test the homogeneity of the alloy. Some samples were taken

from separate members and appendages like covers, handles and decora-

tions, to find out if the separate members came from the same or different

alloy melts. Since restraint had to be used in sampling, and because high

accuracy for the major constituents was not required, the minimum amountof metal taken for wet analysis was about 0.2 grams (200 mg.), and an addi-

tional 0.02 grams (20 mg.) or more was taken for spectrometric analysis.

The initial drillings were rejected, especially if they were contaminated with

products from the corrosion layers. All samples were examined micro-

scopically to check for uniformity. Since the bronze alloy of most of these

vessels is brittle and short, the drillings are finely granular. It was found

that even high-speed steel drills dulled quickly, so to avoid putting too much103Data relating to constancy of composition will be presented in Chapter V and Appendix I.

35

Figure 12

Figure 13

Figures 11-14. The sampling of the bronzes. Figure 11.

The simple equipment used in taking samples from a

bronze vessel was by drilling. The drill is rotated in a

flexible shaft driven by a 1/15 horsepower motor. Themotor speed control device (right) permits use of the drill

at low speeds without loss of power. A foot switch (not

shown) allows additional precision control of speed. Sur-

rounding the vessel are pads and lead shot bags for pro-

tecting and supporting the vessel at any convenient angle;

a square of glazed paper to catch the drillings; a ring-stand

to support the paper; tared weighing dishes; a cuppedslide with cover for storing the sample. Figure 12. This

4

Figure 14

view shows how the vessel is protected and supported

during the sampling operation. Figure 13. Samples for

chemical analysis can also be taken with a small hand-

operated drill. The sample is usually taken from hidden

places like the underside of a leg or the lower end of a

flange. During drilling the vessel is protected with blankets,

sponge rubber pads, and lead shot bags. Figure 14. Thedrillings for analysis are transferred to a weighed micro-

culture (cupped) slide. The hole left by the drilling is

plugged and concealed with a paste made from dry pig-

ments and polyvinyl acetate emulsion. Pigmented wax can

also be used.

36

Figure 1

5

Figures 15 and 16. The major constituents of the bronze

samples were determined by wet chemical analysis.

Figure 15. The tin is separated and weighed as tin oxide.

Here the operator (Mrs. L V. Bene) is filtering off the tin

oxide to separate it from soluble copper and lead salts which

remain in the filtrate. Figure 16. The copper is determined

Figure 16

by depositing it electrolytically on a platinum cathode; lead

is deposited as lead dioxide on the anode. All determina-

tions are made in duplicate. A rectifier for providing direct

current, an instrument panel and a variable transformer

for controlling the speed of the motor stirrers are shown in

the background.

Figure 17

Minor and trace elements in the bronze samples were spectrograph at the Conservation Center, Institute of Finedetermined spectrometrically. Mrs. Elisabeth Fitzhugh is Arts, New York University. (Courtesy, Conservationmeasuring a spectrogram taken on a 3-meter Jarrell-Ash Center.)

37

pressure on the object while sampling the drills were replaced frequently.

The problem involved in sampling of ancient metal objects has been re-

viewed by R. M. Organ who has reiterated the view long held by metal-

lurgists that alloy mixtures of metals held at ambient temperatures for long

periods, even over many thousands of years, undergo no change.104 Themany factors involved in the attempt to get a fair sample from a metal

antiquity have only recently been reviewed by Professor Earle R. Caley.105

On some bronze antiquities corrosion has gone so far that no fair sample of

original metal is available.

Since the holes made by the sampling drills can be plugged and effectively

concealed, they in no way detract from the aesthetic appeal of the objects

{fig. 14). The objects become no less valuable for exhibition purposes but

immensely more valuable to scholars because of the knowledge gained from

the sample taken.

Wet chemical analysis

The 200 mg. sample was divided into two parts to permit duplicate

analyses which were carried out by classical gravimetric and electrolytic

methods.106 Each 100 mg. sample was dissolved in nitric acid; tin was

separated as metastannic acid and determined as stannic oxide {fig. 15), and

from the filtrate copper and lead were separated and determined simulta-

neously by electrolysis (fig. 16). When the total of copper, tin and lead was

below 97.5 percent, iron and nickel, if present in more than trace quantities,

were also determined by wet methods; iron was precipitated as ferric

hydroxide and weighed as ferric oxide, and nickel was precipitated and

weighed as nickel dimethylglyoximate. In bronzes of later date zinc was also

sometimes estimated by wet analysis. Here the zinc, after removal of tin,

copper and lead, was precipitated with diammonium phosphate and weighed

as zinc pyrophosphate. Iron and nickel (and zinc) have only been reported

when the concentration exceeds 1 percent; below this the amount of precipi-

tate to be weighed is too small to give accurate results.

All analyses were done in duplicate to test the precision and constancy of

operation and to assure confidence in the results. The figures given for per-

centage composition in Table IA are the averages of the duplicate analyses.

Because of the small sample size, and the non-homogeneity of the metal,

percentage values are rounded off to the first decimal place since only the

first decimal figure is regarded as significant (see Appendix I).

The working time required for a wet analysis including taking of the

104 "Are analyses . . . ?," p. 54.105Analysis . .

. , Chapter 2.106Where lead was present in amount less than 1 percent, it was estimated spectrometrically.

38

sample, sample preparation, analysis and calculations is about eight hours.

In analytical work it is usually more economical of time to carry along

several analyses simultaneously.

Spectrometries analysis

The quantities of minor and trace constituents of the alloy were estimated

by emission spectrometric analysis. This method was employed first, because

of the small size of the metal sample available which did not permit analyses

of trace elements by wet methods; second, because of the high sensitivity of

the spectrometric method for trace quantities; and last, because of the saving

in time and labor as compared with wet analytical methods.

The actual analyses were done at Brookhaven National Laboratory,

Upton, Long Island, New York, and in the later stages at the Conservation

Center of the Institute of Fine Arts, New York University, where, in both

places, a 3-meter Jarrell-Ash grating spectrograph was put at the disposal of

the Freer Laboratory staff (fig. 17). The carbon-arc method of excitation,

using direct current, was employed, and the samples were burned by stan-

dard procedure. Exactly 10 mg. of the metal samples, or the equivalent of

10 mg. of metal in the case of the standard, were weighed into the cup of a

preformed carbon electrode (United Carbon Products Co. Type 1334 and

Type 101-L). The cassette was centered at about 5800A so that the spectra

recorded in the second order ultra-violet were in the range from 2200A to

3500A. The slit height was 2 mm. and width 80^, and a 23 percent trans-

mission screen filter was placed between the slit and the sample. Each

sample was burned to completion in H to 2 minutes in an arc adjusted to

8-9 amperes at about 300 volts. Ilford Line photographic film, on an acetate

base was used ; it was developed in D- 1 9 developer for four minutes.

The Freer bronze samples were arced in duplicate, on separate plates, to

eliminate any difficulty due to possible loss of sample. In respect to the sam-

ples analyzed at New York University, conditions and method varied

slightly. No filter was used, and the slit width was reduced to 20/x.107 Thevoltage was about 200. The film used was Kodak Commercial PB film, onpolyester base.

Preparation ofstandardsfor spectrometric analysis

It was realized that various copper alloy standards from the National

107A wide slit (80/n) was used at Brookhaven because it was thought that small differences in concentrationwould show up better in the wide lines than in narrow ones for purposes of visual comparison. Later workshowed that the use of a narrow slit, such as 20/u not only produces a more background-free spectrum, but it

also is equally easy to use visually, within 50 percent range of error. It was also found that determination ofAs and Sb was possible at a lower concentration with narrow lines than wide ones. New standard plates wereprepared under the different conditions.

39

Bureau of Standards and other sources could not easily be used directly as

standards for visual comparison with the Freer bronzes because those

standards would not supply a graded series of different concentrations of the

various elements. It had also been found that standards of isolated elements

were difficult to use for visual comparison because they have few lines to

serve as reference points. The unknowns were to be compared to the stan-

dards visually since the determination of all elements with the aid of a

densitometer would have been a time-consuming process not justified in this

particular project.

It was decided to use standards with copper oxide as matrix, with other

metal oxides added. Preliminary experiments indicated that it was valid to

compare metal oxide mixtures with metal alloys. The copper oxide matrix

standards were prepared by mixing pure copper oxide with Spex powder, a

commercially available spectrographic standard made up of oxides of 43

elements, each in 1.43 percent concentration. A range of 10 concentrations

of minor elements from 0.001 to 1.0 percent was used to prepare the stan-

dard plates. These standard plates were compared with plates made from

National Bureau of Standards and other copper alloys to determine their

accuracy. It was found that the error was never higher than 50 percent of

the reported value.

Estimating quantities presentfrom spectrometric films

The visual comparison of the unknowns with the standards was done

usually with the aid of a light box and stereoscopic microscope. The lines

were inspected and compared at approximately x 6 magnification. In order

to facilitate visual comparison, more space was left horizontally between

the spectra than would have been done ordinarily.

The spectrometric results in Table IB are reported as percentage values,

and not within certain ranges, as is sometimes done. These figures are valid

within a range of 50 percent of the reported value; i.e., a concentration

reported as 0.05 percent might range between 0.025 and 0.075 percent.

Hence, none of the figures listed in the spectrometric tables is to be taken

as a precise and exact amount of the element present, but as "of the order

of;'

The tables and charts

The major constituents of 121 bronzes determined by wet methods are

reported in Table IA, and the minor and trace constituents estimated

spectrometrically are listed in Table IB. Nineteen of these vessels which are

classed as "recent" are indicated with a single asterisk. The composition

40

data from these tagged bronzes are not included in the data used to compute

averages and to construct the charts showing distribution of major or minor

constituents. They are not used in special Tables II to V. The data on the

tagged bronzes, however, are incorporated in Tables IA and IB for con-

venience in finding.

Discussion ofthe analytical data: Major constituents

A glance at the tables of analyses shows that the ceremonial vessels are

made of copper-tin alloy with widely varying amounts of lead. These three

elements make up, in most cases better than 95 percent of the total composi-

tion of the vessels and in many instances, better than 97.5 percent. In only

five of the bronzes is the total of major constituents between 90 and 95

percent; where the total of major constituents is below 97.5 percent it is

presumed that the metal is deeply corroded and hence the sample was

contaminated with oxygen mostly in the form of cuprous oxide. Three of

the bronzes were so deeply corroded that no fair sample could be taken.

Minor and trace elements account for only 1 to 2 percent of the total

content. Except where excess lead (over 4 percent) is used, the alloy of manyvessels is good quality bronze.

Copper

Copper is the principal metal. It varies from a maximum of 93.3 percent

(tou 39.41) to a minimum of 63.7 (chien 15.107), a range of 29.6 percent.

The histogram for copper is shown in Chart 1 {fig. 18). This element

accounts for the reddish to yellowish color of the alloy and for the green,

blue and red corrosion products on the surface. It accounts for the strength

and toughness of the alloy, and also for its high melting point.

Tin

Tin is the next most important constituent. Its content ranges from a

high of 21.5 percent (fang-i 30.54) to a low of 1.7 percent (tou 39.41), or a

range of 19.9 percent. Tin content of over half of the bronzes falls in the

range 13 to 17 percent, as shown in the histogram for tin (Chart 1). This

metal was no doubt added intentionally to lower the melting point of the

copper and to make the melt more fluid so that, when cast, it poured easily

and filled the details of the mold. The presence of tin affects the color of

the alloy in proportion to the amount present; as the tin content increases,

the reddish color of copper is changed to yellow, like brass, and in amounts

upwards of 20 percent, the alloy becomes steely in tone. As the tin content

of bronze rises, the brittleness of the alloy increases so that in the vicinity of

41

about 25 percent tin it becomes quite useless except for reflecting purposes,

as in a mirror. The proportion of tin not only governs the color of the alloy,

but to some extent also affects the color of the patina.

LeadLead varies widely from a high of 26.0 percent (chien 15.107) to a low of

0.007 percent (tsun 1 1 .40) with all values between. There is less lead than

tin in 68 (about two thirds) of the vessels analyzed. The lead content of

48 of the vessels is 5 percent or less, and in 12 of them lead is 1 percent or

less. The histogram for lead is shown in Chart I. The alloy in most cases is

"leaded bronze. " It is recognized that lead in bronze up to about four per-

cent is taken into solution in the melt, but above that amount it separates

on slow cooling, and the lead sinks to the bottom of the crucible or mold.108

Hyman describes how lead above 4 to 5 percent is distributed throughout

the melt as tiny lead droplets similar to the oil droplets in an oil-in-water

emulsion.109 These droplets are trapped throughout the bronze more or less

homogeneously, for the most part, between crystal grains if the melt is cooled

rapidly. Chikashige says that lead up to 4 percent, taken into solid solution,

further helps to reduce the solidification temperature of the melt and makes

it more fluid. Lead scattered throughout the hardened alloy makes the

bronze softer, and hence, easier to tool and easier to finish. It is probable

that lead was used intentionally to facilitate the casting of the bronze, but it

also may have been used as an adulterant and substitute for the more

costly tin. Chikashige says, however, "It [lead] can not be taken as an im-

purity, although it does not appear in the ancient recipe [of the Kao KungChi] and is besides detrimental in bronze-casting owing to its tendency to

liquation." 110 It may have been recognized that lead could be generously

used in vessels where, unlike weapons and utensils, strength was not a prime

consideration. There is no evidence in our data that Shang bronzes differed

from Chou bronzes in respect to lead content. Lead is erratic in both classes.

Like tin, lead in excess contributes to the color of the patina on a bronze

because, during corrosion, the lead is converted mainly to white lead car-

bonate or cerussite which, along with tin oxide, causes the pale green tone or

so-called "water patina." It is also obvious that bronze alloys which contain

excess lead, especially those with a lead content exceeding their tin content,

do not form a desirable patina. During burial they tend to acquire thick and

crusty grayish alteration products mostly of carbonates and oxides of lead,

which are disfiguring and without aesthetic appeal.

108See phase equilibrium diagram for Cu-Pb, Chapter V.109"The function of lead . . .

."

110Oriental alchemy, pp. 77-82.

42

Minor and trace constituents

Except in a few instances, all other chemical elements present are in quan-

tities estimated to be below 1 percent and most of them are below 0.1 per-

cent. All of these minor and trace elements were probably introduced as

impurities in the three major constituents, copper, tin and lead. In about

three-quarters of the spectrometric analyses reported, the estimate for total

impurities is below 1 percent. By the methods of analysis used, only ten

elements were regularly detected in the group classed as impurities. Four

other elements were often but not regularly detected in trace amounts. The

distribution of these elements is shown graphically in Chart 2 {fig. 19).

Iron

Iron is the most abundant impurity, averaging (in 100 analyses) 0.2 per-

cent. This element could have come from various sources, the original

copper, lead and tin ores, the smelting agents, and the crucibles. Only one

bronze (yu 47.12) has iron content above 1 percent, and in this case there is

enough iron in the bronze to give the patina of the vessel a yellowish tone.

Silver and gold

Next to iron, silver is the most abundant impurity and averages about 0.1

percent for all the samples analyzed. It probably came in with both copper

and lead because it is a common impurity in the ores of both metals. There

is not enough silver to affect the metallurgy, the properties of the alloy, or

the nature of the patina. Gold is absent except in trace quantities in about

four-fifths of the total number of pieces.

Nickel and cobalt

Both elements are found as impurities in copper ores, and they probably

pass into the reduced copper with little alteration in ratio to the copper or in

ratio to each other. Nickel in 100 bronzes averages 0.032 percent; cobalt

0.013 percent. It is interesting that the ratio of nickel to cobalt in these

alloys is about 2:1 which is higher for cobalt than it is in nature. The general

consensus of opinion among geochemists based on data collected by

Fleischer is that the ratio of nickel to cobalt in the earth's crust is about

7:1.111

Zinc

This is the mystery element of the series. Zinc is totally lacking even in

spectrometrically detectable quantities in about 85 percent of all the bronzes111 U.S. Geological Survey Circular No. 285, pp. 4-5.

43

examined. It occurs in 6 of the 100 vessels dated from the Shang through the

Han dynasties but only in amounts below 0.1 percent. In 9 bronzes the zinc

content ranges from 0.1 to 3.7 percent {tsun 09.257), but all these are of

uncertain date, perhaps Sung dynasty or later bronzes modeled in archaistic

style, or outright fakes. None of the alloys analyzed has enough zinc to be

called brass; the zinc was introduced probably not by intent but by accident

(see Chapter II). All the bronzes that contain zinc also contain tin and lead.

Apparently sometime during or after the Han dynasty zinc gradually edged

its way into bronze alloy composition, and this is confirmed by the analyses

of post-Han bronzes shown here. Many more analyses will have to be made,

especially on post-Han bronzes, before the history of zinc in China will begin

to unfold.

Arsenic, antimony and bismuth

The first two elements are mostly in the range of 1.0 to 0.01 percent and

in about the same relative abundance as in nature. In only two bronzes

(kuang, 49.10 and fang-i, 15.136) is the arsenic content high enough (1 per-

cent) to indicate that arsenical copper ores were employed. In a quarter of

the bronzes, arsenic was not detected; and in many it is hardly more than a

trace impurity. Bismuth seems to be a persistent heavy metal trace impurity

averaging 0.04 percent in 100 bronzes, which is below silver. It may be

present in sufficient quantity, however, to increase slightly the brittleness of

the bronze. The anitmony content is erratic. In five of the bronze samples

analyzed, antimony was estimated to be present in quantities of 1 percent or

greater; and in all of these five bronzes the content of lead exceeds that of

tin. The wet analysis estimate for tin in the antimony-bearing bronzes is

probably slightly in error (too high) because, in the decomposition of the

sample by nitric acid, antimony oxide precipitates quantitatively with the

tin oxide and hence the two oxides are weighed together.

Silicon

As would be expected, this element shows up persistently in the 0.01 to

0.1 percent range. The average content of 0.06 percent is about twice that of

nickel.

Other elements

These were detectable in trace amounts, some consistently, others spora-

dically. Some like calcium and magnesium appear to have their source not

in the bronze sample but as impurities in the carbon electrodes used to

support the samples during spectrometric analysis. Their amounts and lower

44

limits of detection are shown in Table IB notes. Also, listed in Table IB

notes, are a number of elements that were sought for but not found.

In the original planning of the project, no special consideration was given

to the identification and estimation of phosphorous. In the routine examina-

tion of the spectrograms, no phosphorous lines were recognized. It is un-

fortunate that the strongest phosphorous line (2535.65A) is masked by

iron (2535.604A) which is present in all of the bronzes sampled. The other

strong phosphorous lines (2553.28, 2554.93, and 2534.01 A) do not show in

the spectrograms of any of the bronze samples analyzed, but they are not

present in the standards employed in concentrations of 1 percent or less.

This negative information is supplied here because, since the beginning of

the project, questions have been raised about the possible phosphorous con-

tent of ancient bronzes. Professor Cyril S. Smith has pointed out that the

element phosphorous has an efTect on bronze casting far out of proportion

to the small amounts in which it may be present.112 Phosphorous acts as a

deoxidizer and makes the bronze melt more fluid, resulting in a better cast-

ing. Where phosphorous does occur in early metallurgy, its source was

probably bone ash, introduced as a flux.

Relationship between composition and time oforigin of the vessels

The question is frequently asked: Do the vessels of different time periods

show any difference in average composition in respect to the major elements,

or even the minor elements? Close inspection of the analytical results showthat, on the whole, they do not. There seems little or no relationship be-

tween composition and age. In Table II are grouped all of the bronzes by

periods from Shang to Han. Comparison of the average composition of

copper, tin and lead in each group shows no consistent progression. The

average of lead content for middle-late Chou bronzes is slightly higher

than the average for earlier and later periods, but this appears to be purely

accidental. It is not caused by the large number of crudely made vessels in

this group called ming-ctii (Table V), which were made solely for burial

purposes. The data in Tables III and IV show no relationship between

composition and types of vessels, nor is there any steady increase or de-

crease of any element in any one type vessel over a period of centuries. Acase in point is provided by the two fang-ting 50.7 and 09.261 (see Vol. [, Nos.

34 and 35). Their major element compositions differ by only one percent

which is well within the confidence limits, and the spectrometric analyses

of the two are close, though not identical. One difference is that zinc is

present in 09.261, though in a quantity less than 0.03 percent; 09.261 has

112"Materials and the development of civilization and science," p. 917, note 10.

45

been called "recent" and "an imitation, " while 50.7 dates from the early

Chou period. Yet it would be very difficult to tell the two vessels apart on

the basis of composition. Hence, a comment by W. Perceval Yetts madesome years ago (1931) on the article by W. F. Collins published the same

year is still valid

:

"Unless the exact age of an object can be determined, the attempt is

futile to relate its metallic composition with a definite school. Moreover,

the question of provenance is of the utmost importance. Beyond doubt

there existed at the same time in ancient China different local centres

of bronze casting. Probably the practice as regards the composition of the

alloy varied as much as their respective standards of style and craftsman-

ship. To imagine that bronze casters in many of the numerous semi-iso-

lated states which constituted ancient China were governed by a uniform

rule is, of course, illusory. Their practice was probably as diverse in

regard to material as in other respects. Subject to certain rough empirical

standards, they must have used the metal which was obtainable. Presum-

ably they put into the melting-pot every scrap that came to hand, muchin the same way that many casters do at the present day." 113

Among the minor elements interest centers chiefly on zinc. It has already

been pointed out that bronzes of the early periods contain no detectable

zinc, and those few bronzes in which zinc is just detectable are late Chou or

Han. It is believed that in post-Han times the occurrence of zinc becomes

more and more frequent, but as yet there is little data on which to base this

assumption. In order to come to firm conclusions in respect to relationship

of composition to chronology, we have to have the analyses of not a few

dozens of bronzes but many hundreds - even better, thousands.

Relationship between composition andgeographic origin

There is none from the data presented here because the exact place of

origin of none of the Freer bronzes is known. The same is true of the bronzes

listed in Appendix II. In general, there is insufficient knowledge of mines

and mining in ancient China to attempt to relate composition and ore

sources.

1 1 3 yetts, "Problems . . .," p. 2; comment on Collins, "The corrosion of .

.

46

«0

N

Ok

o

50

46

42

38

34

30

26

22

18

14

10

6

2

Copper Tin Lead

60 65 70 75 80 85 90 95% 1 5 10 15 20 25% O 1 5 10 15 20 25 30%Copper content Tin content Lead content

Figure 18

Chart 1 . Distribution of major alloy constituents in 97 vessels. Data from wet chemical analysis.

-k

k

-k.

O

5.0

/

09

05

OJ009

005

OO/

Not ADetected V

Ag Au Fe Co Ni As Sb Bi Zn Si Cr Al Mg Mn

Figure 19

Chart 2. Distribution of minor and trace elements in 100 bronze ceremonial vessels. Data from spectro-

metric analysis.

47

TABLE IA

Major constituents of bronze ceremonial vessels deter-

mined by wet chemical analysis (with confidence limits)

FGA Type Lu (CL an (LL rb (LL Other 1 otal Sn/XTNo. a / 95% ) % 95% )

o/% 95% ) % % rb

94.17 kuei 82.6 (±1.9) 10.2 (±4.0) 3.5 (±.25) 96.3 2.9

07.33 lei 66.1 (±•5) 5.1 (+4.5) 24.1 (+2.16) 95.3 0.2

07.34 ku 76.7 (±1.27) 15.2 (+3.3) 4.9 (+.51) 96.8 3.1

07.37 chia 82.3 (±3.2) 12.2 (+ 1.65) 1.0 (±•7) 95.5 12.2

09.254* huo 68.9 (+2.48) 3.8 (+.25) 21.8 (±•51) Zn3.5 98.0 0.2

09.257* lei 74.5 (±6.23) 5.1 (±7.19) 15.0 (±•32) Zn3.7 98.3 0.3

09.258 yu 79.6 (±2.5) 16.4 (+3.94) 1.9 (+.38) 97.9 8.6

09.259* kuei 71.9 (+5.28) 7.9 (± 1.65 17.3 ( + 2.54) 97.1 0.5

09.260* yu 79.7 (±•45) 14.0 (± 1.08) 1.2 (+.32) 94.9 7.8

09.261* fang-ting 78.7 (±1.59) 13.9 (+ 1.46) 4.7 (±.83) 97.3 2.9

09.279* tsun 67.2 (±5.02) 8.0 (+.06) 22.0 (+2.29) 97.2 0.4

09.280 p'ou 79.4 (±1.65) 8.9 (±•13) 7.6 (±1.5) 95.9 1.2

09.333* ting 84.6 (±•57) 3.5 (±11.38) 7.9 (±2.23) Zn2.3 98.3 0.4

09.334 p'ou 71 .7t — 12.2| —13.9f

— 97.8 0.9

09.335§ p'ien-hu 66.9 (±1.21) 8.6 (+1.97) 21.9 (+.44) 97.4 0.4

09.336 f" 64.7 (±1.14) 9.2 (±.95) 23.8 (±0) 97.7 0.4

11.35 p'an 71.6 (±2.6) 8.7 (+6.4) 15.0 (+3.4) 95.3 0.6

11.36* yu 66.3 (+13.67) 9.0 (+2.04) 22.1 (+1-14) 97.4 0.4

11.37* yu 74.5 (±7.12) 12.4 (+.64) T5.4 (±•13) 97.3 1.2

11.38 kuei 76.0 (±5.9) 14.8 (±8.2) 7.9 (± 1-2) 98.7 1.9

11.39* chiieh 83.5f— 13.4| —

%— 96.9 —

11.40 tsun 82.2 (+8.01) 14.8 (+3.0) — — 97.0 —11.41* ting

p'an

72.7 (±2.35) 17.9 ( + 3.12) 9.4 (+13.17) 100.0 1.9

11.44 72.5 (±1.02) 7.9 (± 5.34) 16.8 (±2.04) 97.2 0.4

11.45 an t— t

+— t

4-— — —

11.46 huo 75.3 (+2.86) 9.9 (+ 1.65) 9.7 ( + 4.13) 94.9 1.2

11.47 huo 73.0 ( + 9.48) 12.0 (+ 18.4) 10.8 ( + 7.44) 95.8 1.1

11.49* kuei 68.4 ( + 4.64) 5.9 (+4.45) 22.0 (+.64) Zn 1.5 97.8 0.3

11.50* vu 78.0 (±.70) 6.9 (+.45) 11.2 (+.38) 96.1 0.6

11.51 'ku 64.0 (±5.72) 5.9 (±7.5) 25.9 (±•5) 95.8 0.2

11.53 kuei 82.6 (±2.54) 13.2 (+1.65) 1.5 (+1.53) 97.3 8.8

11.54* kuei 85.6 (±•64) 11.0 (+.64) 0.2 (+1.72) 96.8 55.0

11.55* vu 71.7 (±•83) 10.4 (±1.40) 13.7 (±1.59) 95.8 0.7

11.56* hu X— +

+t+

— —11.57 p'ou 80.1 ( + 3.50) 8.3 (±4.39) 3.8 (+.06) 92.2 2.2

11.58 kuei 72.1 (±•32) 11.8 (±12.66) 13.4 (+.89) 97.3 0.9

11.59 hu 77.3 (±2.73) 12.4 (+1.97) 8.4 (+1.21) 98.1 1.5

11.60 ting 76.5 (±2.16) 6.4 (+1.27) 15.3 (+.70) 98.2 0.4

11.66* deep tray 77.4 (±2.42) 10.1 (±•13) 9.8 (±.13) 97.3 1.0

11.81* tui 69.2 (±•51) 5.6 (+5.98) 21.5 (+1.78) Zn0.9 97.2 0.3

11.82* yu 70.4 (+2.86) 4.9 (+8.40) 19.1 (+ .06) Zn3.5 97.9 0.3

11.630 huo 72.8 (+4.20) 7.8 (+2.48) 18.0 (+.06) 98.6 0.4

12.72* kuang 73.4 (+5.53) 7.7 (+.70) 15.8 (±11.89) Zn2.2 99.1 0.5

13.14 fu 67.8 (±•70) 7.0 (±.51) 22.0 ( + 2.80) 96.8 0.3

13.21 hu 80.8 (+2.10) 10.8 (+1.08) 3.9 (± -70) 95.5 2.8

13.30* p'ou 81.9 (±•51) 17.8 (+.06) 0.0 99.7

15.102 kuei 77.2 (+4.20) 16.2 (+6.93) 3.2 (±0) 96.6 5.1

15.103 pien-hu 81.8 (±1.72) 2.5 (+3.05) 11.7 (±•45) 96.0 0.2

48

TABLE IB

Minor and trace constituents of bronze cere-

monial vessels estimated by emission spectrometry

Ag%

Au Fe Co Ni As Sb Bi Zn Si Other FGA% % % % % % % % % % No.

.2 nd .2 .004 .02 2 .2 .03 nd .02 94.17

.2 nd .3 .03 .02 .3 .03 .03 nd .03 07.33

.2 nd .1 .001 .009 .3 .03 .09 nd .02 07.34

.01 nd .01 .001 .001 nd nd nd nd .07 07.37

.2 nd .1 .002 .09 .3 > 1.0 .05 > 1.0 nd 09.254*

.1 .01 .2 .003 2 2 .5 .03 > 1.0 <.001 09.257*

.07 nd 3 .007 m 2 .1 .04 nd .01 09.258

.2 nd .01 .005 .02 .07 .07 .2 nd nd 09.259*

.09 nd .3 .003 .03 .07 .07 nd nd .01 Pb .4 09.260*

.2 nd .3 .005 .03 .3 .07 .07 <.03 .02 09.261*

.1 nd .2 .003 2 .3 .5 .07 .1 < .001 09.279*

.1 nd .07 .001 !009 .3 nd nd nd .1 09.280

.1 nd .3 .001 .03 .3 .2 nd 1.0 .04 09.333*

.02 nd .02 < .001 .001 nd nd nd nd .002 09.334

09.335§

.07 nd .09 .1 .05 nd .02 nd nd .009 09.336

.2 nd .3 .01 .02 .2 2 .05 nd .001 11.35

.2 <.01 .07 .003 .03 2 A .05 nd nd 11.36*

.1 <.01 .1 .007 .02 A .07 .09 nd .001 11.37*

.2 nd .06 .007 .03 .3 .1 .07 nd .01 11.38

.09 nd .3 .001 .02 <.07 nd .03 nd .005 Pb .03 11.39*

.1 <.01 .2 .001 .02 .2 .05 .09 nd .002 Pb .07 11.40

.09 nd !03 .02 .03 .1 .03 .03 .05 .005 11.41*

.1 <.01 .02 .02 .03 .1 .3 .03 nd .03 11.44

.2 <.01 .2 .03 .03 .3 .07 2 nd .04 11.45

.2 <.01 .01 .01 .03 .1 .3 .03 nd .. .05 11.46

.2 .01 .02 .007 .03 .2 .09 .05 nd 1.0 11.47

.1 nd .09 .003 .2 .3 .5 .03 > 1.0 nd 11.49*

.1 nd .2 .01 !05 .2 .01 <.03 nd 1.0 11.50*

.2 nd .3 .02 .02 .1 .5 .2 <.03 .01 11.51

.1 nd .2 .01 .02 .3 .1 .2 <.03 .001 11.53

.09 nd .3 .01 .02 nd nd .2 .03 .02 Pb.2 11.54*

.1 <.01 .7 .01 .01 .2 .1 .2 nd .001 11.55*

1 1.56*

.2 .01 .01 .002 .02 .1 .2 .03 nd 2 11.57

.2 <.01 .01 .002 .03 .2 .2 .03 nd < .001 11.58

.2 <.01 .03 .002 .02 .05'.2 .03 nd > 1.0 11.59

.2 .01 .003 <.001 .02 nd .2 .03 nd .07 11.60

.2 nd .003 .01 2 .1 3 .03 .1 <.001 11.66*

.2 nd .07 .007 .07 .2 2 .1 1.0 nd 11.81*

.2 nd .2 .01 .1'.2 2 .03 > 1.0 .001 1 1.82*

.2 nd '.01 .003 .03 .2 A .03 nd .002 11.630

.2 nd .1 .002 .09 2 .2 .07 > 1.0 nd 12.72*

.2 nd .1 .03 .03 A .02 .05 nd .04 13.14

.2 nd 1.0 .01 .02 .05 .03 .03 .03 .02 13.21

.03 nd .2 .001 .002 <.07 nd nd nd <.001 Pb.3 13.30*

.07 nd !07 .002 .02 <.07 .03 .03 nd .04 15.102

.1 nd .006 .06 .1 nd .005 .2 nd .002 15.103

49

TABLE IA—Continued

Major constituents of bronze ceremonial vessels deter-

mined by wet chemical analysis (with confidence limits)

FGA Type Cu (CL Sn (CL Pb (CL Other Total Sn/No. 0/ 95%) % 95%) 0/ 95%) % 0/ Pb

15.104 lei 64.7 (+4 26) 6.1 (+ .25) 25.9 ^+7.38) 96 7 9

15.107 chien 63.7 (±8.33) 7.9 (+ 13) 26.0 ^+8.14) 97 6y l . vj U.J15.136 fcuw-i 67.8 ( + 0) 7.1 (+3.82) 21.9 lJ+3.24) 96 8 3

16.142 tsim 79.4 (+ 95) 13.0 (±4.90) 5.0 I[± .25) 97 4 9 6

16.248 huo 87.1 (+ 1 27) 5.3f 3.4 ;+ 1.40) 95 8 1 6

16.361 83.3 (+2.16) 14.5 (+ 5.09) 0.0 97 8

16.480 kuei 78.2 (+ 38) (17 1) (+ 1.78) 3.2 i{+ 1.46) 98 5y o . j S 3

17.193 kuei 78.5 (+ 25) 10.3 (± .06) 8.5 '+ 19) 97 3 1 9

17.202 ku 76.7f 19 7t 1 6t 98 6 1

19.6 chih 77.5 ( + 1 97) 14.2 (±2.61) 4.3 [± .06) 96 3 3

23.1 cliia 80.0 ( + 9 5) 14.3 (+ 1? D 4.7 + 13) 99y y ,\j 3

23.2 lien 78.8 (+4.01) 5.7 (+4.13) 12.1 + 1.14) 96 6 5V/. .J

24.1

1

kuei 65.8 (±•32) 7.2 ( + 6.68) 22.8 + 1 721 95.8 3\y

.

24.12 hu 73.0 (+3.82) 9.8 (+ 1 721 15.6 f+ 3~>) 98 4 6

24.13 tui 85.9 (+2.29) 4.1 (+3.82) 7.1 ( + .06) 97 1y i . i 6

24.14 kuei 82.0 (±3.37) 13.8 ( + 9.03) 1.8 !±.89 ) 97 6 7 7

25.2 fnnv-t sun 79.7 (±3.75) 13.6 (±5.47) 2.0 ± .76) 95 3 6 8

25.3 chileh 80.5 (±2.35) 12.7 (+ 1.91) 5.0 i± 1-65) 98.2 5 430 96 Vlly Li 67.4 (+4.01) 14.0 (+ 1 72) 13.4 '+

1 08) 94 8 1 1

30.54 J u"b ' 77.7 ( + 4.20) 21.5 (+ 1.46) 1.2 (± 1.59) 100 4 1 8

31 10 kupiAHCI 70.3 (± 10.56) 10.0 (± .83) 14.7 '+11 45) 95 7V/. /

32.13 tun 70 2t' J --

1 14.3f 1 3.0f 97 5 1 1

33 2 hi in 71.1 (+3.56) 15.7 (+ 57) 12.3 99 1 1 3

35.12 chici 75.1 (±5.79) 15.6 (+2 93) 9.5 HO) 100.2 1.6

36 6 hiin t %37 1 Vllyu 84 1 (±1.21) 13.3 (+ .38) 1 .2 (+ .06) 98 6 1 1

38 5•JO • *J leufin&l\ if Li 1

1

C 69 1v./ y . i (±6.17) 15.1 (+6.23) 1 3 9 Mil 77) 98 1y o • i 1 11*1

38 6 chihVsl III I 78 8 (±1.53) 15.1 (± .83) 3.4 (+2 80) 97 3 4 5

38 7 tunl til l 74 3 (±.64) 12.0 (+ 1 (P) 8 9 (+ 1 97) 95 9 1 4

38 9f) Jrupirv if C t8"> (±.70) 13.2 (+ 57) 3.6 (+ .76) 98 8 3 7

39 5 phipn 70 8 ( + 4.96) 12.8 (+ 1.53) 14.7 ( + 3.82) 98 3V *J . *y 1

39.41 ton 93.3 (±14.25) 1.7 (± .06) 2.3 (±.70) 97.3 0.7

39 53 f\ It til ICL 70 It 11.6f 15.6f 97 3y i »<y 0.7

40 3 ku 75 9 (±3.12) 19.0 ( + 2.48) 4.8 (+ .25) 99.0 4.0

40 1 1 Vllyu 89 8 (±•57) 14.6 (+ 1.46) 3.7 (+ 1.59) 101.1 8.3

41 8 If 11PIt\ Lic / 74 9 (±•76) 14.3 (+ 1~>7) 10 (+ 19) 99 9y y

,

1.4

49 1 nun 78 4 (±5.15) 13.6 (+ 5.34) 3.1 (+ 45) 95.1 4.449 14 Vllyu 74 4 (±2.73) 15.7 (+ 1 65} 7 8 (+] 91 ) 97 9y i • y 2.0

43.9 'ku 73.8 (+3.82) 17.5 (+.38) 7.0 (+.51) 98.3 2^5

44.1 tsun 78.4 (±•32) 17.2 (+ 1.46) 0.2 (+.06) 95.8 86.0

46.4 ting 76.0 (±.76) 16.0 (±5.41) 6.0 (±1.65) 98.0 2.7

46.11 lien t% t

46.31 ting 86.0 (+2.29) 9.7 ( + 2.48) 0.4 (+.25) 96.1 24.0

47.11 ting 79.6 ( + 4.64) 13.8 (+3.18) 3.1 (+2.61) 96.5 4.5

47.12 yu 74.7 ( + 6.36) 12.6 (+4.13) 5.4 (+2.16) Fe2.5 92.7 2.3

47.20 ting 74.8 (+10.94) 13.7 (+.38) 10.0 (+1.34) 98.5 1.4

48.1 hu 76.4 (±16.54) 18.1 (+2.04) 3.0 (±0) 97.5 6.0

49.5 hu 73.7 (±•76) 12.4 (+4.71) 12.1 (+.19) 98.2 1.0

50

TABLE IB—Continued

Minor and trace constituents of bronze cere-

monial vessels estimated by emission spectrometry

Ag Au Fe Co Ni As Sb Bi Zn Si Other FGA% % % % % % % % % % % No.

.3 nd .3 .02 .02 .2 .09 .03 nd .004 15.104

.3 < .01 .02 .02 .09 .7 > 1.0 nd nd .004 15.107

.3 nd .7 .02 .02 1.0 .2 .06 .07 .001 15.136

.2 nd .07 .001 .01 .2 .03 .09 nd .01 16.142

.3 < .01 .009 .002 .01 2 .07 .05 nd .07 16.248

.07 nd .1 .001 .02 nd nd nd nd .02 Pb.3 16.361

.2 nd .1 .02 .03 2 .09 .07 nd .02 16.480

.09 nd .3 .007 .03 .3 .3 .07 nd .001 17.193

.07 nd .03 .001 .005 2 .07 nd nd .05 17.202

.09 nd .3 .01 .03 .1 .09 .05 .03 .005 19.6

.07 nd .03 .004 .005 nd nd .08 nd .02 23.1

.2 nd .2 .09 .09 .2 .2 .04 nd .002 23.2

.2 nd .09 .01 .03 .2 .1 nd nd .1 24.11

.2 nd .03 .02 .05 2 .1 .05 nd .05 24.12

.3 < .01 .01 .03 .2 .3 > 1.0 .05 nd .006 24.13

.09 < .01 .03 .003 .03 .2 .01 .03 nd .004 24.14

.1 nd 1.0 .03 .009 .1 nd nd nd .02 25.2

.1 nd .02 .008 .09 2 nd < .03 nd .06 25.3

.08 nd .7 .02 .02 .07 .05 nd nd .06 30.26

.2 nd .09 .005 .03 .2 .2 .04 nd .01 30.54

.1 < .01 > 1.0 .03 .03 .3 .04 .05 nd .05 31.10

.2 nd .03 .02 .05 .3 .2 .03 nd > 1.0 32.13

.07 nd .2 .004 .01 .3 .03 .03 nd .04 33.2

.1 nd .04 .004 .03 nd nd < .03 nd .01 35.12

1• J. nd .03 < .001 .003 nd .02 nd nd .01 36.6

1 < .01 .5 .004 .03 .2 .3 2 nd .01 37.1

nd .4 .01 .01 ]8 .03 < .03 nd .06 38.5

07 nd .06 .005 .02 nd nd < .03 nd .03 38.6

2 nd .03 .001 .02 nd .03 nd nd 1.0 38.7

2 nd .3 .008 .03 .2 .04 .08 nd .03 38.20

2 < .01 .1 .01 .07 .3 .1 < .03 nd .01 39.5

.2 nd .05 .001 .09 nd .03 nd nd < .001 39.41

1 nd .1 .002 .02 .3 .01 < .03 .02 .02 39.53

07 nd1 1 VI !o6 .004 .03 nd nd nd nd .02 40.3

1• 1 nd1 1 VI .09 .008 .01 nd nd .03 nd .003 40.11

07 nd1 1 VI .03 .004 .008 nd .02 .03 nd .05 41.8

03 nd1 1 VI .01 < .001 .001 nd .02 nd nd .07 42.1

\ ndI 1 VI 1.0 .001 .005 nd nd .03 nd < .001 42.14

!08 nd .06 .004 .02 .3 .01 .03 nd .04 43.9

.07 nd .4 .007 .01 .3 .06 nd nd .01 44.1

.09 nd .4 .02 .03 .3 .1 .06 nd .06 46.4

.2 nd .03 .03 .09 .3 > 1.0 .03 nd .07 46.1 1

.1 nd .5 .008 .02 nd .01 <.03 nd .04 46.31

.1 nd .01 .002 .01 nd nd <.03 nd .02 47.11

.09 nd > 1.0 .01 .03 > 1.0 .01 .03 nd .01 47.12

.1 nd .2 .03 .09 .05 .09 nd nd nd 47.20

.02 nd .03 <.001 .<001 nd .04 nd nd nd 48.1

.02 nd .02 <.001 nd nd .005 nd nd .03 49.5

51

TABLE IA—Continued

Major constituents of bronze ceremonial vessels deter-

mined by wet chemical analysis (with confidence limits)

FGA Type Cu (CL Sn (CL Pb (CL Other Total Sn/

No. % 95%) % 95%) % 95%) % % Pb

49.10 kuang 78.0 (+1.21) 19.5 (±•51) 0.4 (±•57) 97.9 49.0

50.7 fang-ting 77.7 (+1.14) 14.9 (+1.46) 5.5 (+.76) 98.1 2.7

50.18 fang- 1sun 75.8 (+.83 19.1 (±3.63) 0.8 (±1.59) 95.7 24.0

51.2 lien 67.4f 10.lt —20.0t

— 97.5 0.5

51.5 lien 87.1 (+21.31) 10.4 (+1.97) 0.0 — 97.5 —51.18 ku 72.4 (±13.17) 13.3 ( + 2.61) 11.3 (+.19) 97.0 1.2

51.19 tsun 77.9 (±4.07) 17.2 (+5.15) 2.4 (+.57) 97.5 7.2

53.83 chio 76.3 (+4.45) 15.3 (+4.64) 6.9 (+ -89) 98.5 2.2

54.13 fang-i 78.1 (+2.86) 12.9 (+1.21) 7.1 (+ -32) 98.1 1.8

54.15 chiieh 82.5 (+4.07) 15.4 ( + 6.92) 0.9 (+.83) 98.8 17.0

54.122 tsun 81.3 (+.45) 15.8 (+7.00) 1.8 (±0) 98.9 8.8

55.1 tsun 71.5 (+4.90) 6.5 (+1.21) 19.5 ( + 3.50) 97.5 0.3

56.15 fang-hu 78.0 (+3.95) 7.2 (+5.15) 12.6 (+1.02) 97.8 0.6

56.19 chiieh 80.8 (±•57) 13.7 (+5.91) 3.5 (±5.22) 98.0 3.9

56.26 p'an 87.2 ( + 2.93) 9.9 (+.25) 0.4t— 97.5 25.0

57.22 hu 68.8 (+1.08) 10.5 ( + 2.04) 18.3 (±5.15) 97.6 0.6

59.14 hu 83.1 (+2.35) 15.1 (±2.67) o.it — 98.3 150.0

59.15 ting 82.1 (+.83) 12.0 (+14.25) 6.1 (+2.80) 100.2 2.0

60.18 ting 82.9 (+2.16) 13.6 (+5.53) 2.8 (±•51) 99.3 4.9

60.19 kuei 81.2 (+3.69) 14.0 (+1.21) 1.4 (+.13) 96.6 10.0

60.20 yu 83.3 (+3.05) 15.1 (±.51) 0.1 (±0) 98.5 151.0

61.30 tsun 74.1 (+1.21) 13.6 (+ 1.46) 10.8 (+2.16) 98.5 1.3

61.31 i 79.7 (+2.42) 17.1 ( + 6.04) 0.7 (+1.08) 97.5 24.0

61.32 fang-hu 78.0 (+1.14) 15.3 (+.06) 2.4 (+.19) 95.7 6.4

61.33 kuang 73.4 (+4.58) 11.3 (+1.72) 12.5 (+3.56) 97.2 0.9

66.14§ hu 66.6 (+2.22) 11.3 (+1.72) 18.8 (+.76) 96.7 0.6

NOTES TO TABLES IA AND IB.

For location of samples see Vol. i.

*Doubtful authenticity, not included in averages.

tSingle analysis.

JNo analysis—insufficient sample available.

§Added later, not included in averages.

52

TABLE IB

Minor and trace constituents of bronze cere-

monial vessels estimated by emission spectrometry

Ag A 11AU re 1^ Mi Ac 3D RiDl ZJ

1

ol utner pr. A

O/ 0/7o

0//o /o

0//o /o /o

0//o

0//o

0//o

0//o IN O.

.07 nd .8 .02 .02 > 1.0 .1 .05 nd .01 Pb .2 49.10

.1 nd .3 .02 .04 .3 .3 < .03 nd .04 50.7

.1 < .01 .7 .005 .01 .3 .3 .2 nd .04 Pb .5 50.18

.1 nd .09 .001 .05 _2 .04 .2 nd .01 51.2

.01 nd .03 .01 .09 nd .02 nd nd nd Pb .1 51.5

.09 nd .06 .01 .01 nd nd < .03 nd .01 51.18

.07 nd i_ .004 .004 nd nd nd nd .01 51.19

.1 nd .5 .01 .03 .3 .1 .06 nd .03 53.83

.09 nd .08 .004 .003 .1 nd < .03 nd .02 54.13

.1 nd 2 .02 .03 .3 _2 .05 nd .03 54.15

.09 nd .6 .01 .03 .2 .3 .08 nd .01 54.122

.3 < .01 .3 .02 .02 .3 .6 .2 nd .01 55.1

.2 nd .009 .03 _2 .1 > 1.0 .03 nd .03 56.15

.05 nd .07 .01 .01 nd .01 .06 nd .03 56.19

.09 nd .4 .01 .01 < .1 nd < .03 nd .02 56.26

.2 nd .1 .03 .07 .5 .2 < .03 nd .002 57.22

.1 nd .02 .01 .04 nd .01 .03 nd .01 59.14

.1 nd .3 .01 .03 .2 .02 .03 nd .1 59.15

.07 <.01 .1 .005 .01 .2 .01 nd nd .001 60.18

.07 <.01 .2 .03 .02 < .1 .01 .2 nd .04 60.19

.07 <.01 !oi .03 .02 .2 .02 ]03 nd .005 Pb .07 60.20

.2 nd .007 .02 .07 .2 .2 <.03 nd nd 61.30

.1 nd .2 .02 .03 nd .01 nd nd .01 61.31

.2 nd .03 .02 .07 .1 .03 nd nd .02 61.32

.1 nd .01 nd .005 nd .01 nd nd <.001 61.33

66.14§

NOTES TO TABLE IB

Pb and Sn are 1% or over unless otherwise indicated.

nd=not detected.

Elements determined spectrometrically fall into 3 groups.

I. Elements reported in the table, most of whichoccur in the majority of the bronzes.

Limits of detectability, percent:

Ag, Fe, Co, Ni, Si 001

Au 005Bi 03

As 1

Sb 03

Zn 1

II. Elements present in trace amounts. The limit of

detectability is .001 % unless otherwise noted.

Al < .001-.05% in 40 bronzes. 37 are .004%or below. There is possibly a faint trace in

other bronzes.

Cr <.001-.004% in 42 bronzes.

Mg < .001-.007% in 106 bronzes. 84 are

.001 % or below.

Mn < .001-.004% in 54 bronzes. 52 are

.001% or below.

Ge < .001—.01 % in 4 bronzes.

Cd very faint trace in some bronzes

(limit of detectability .01 %).

53

III. Elements searched for and not found.

Limits of detectability, percent:

B .001 (searched for in 35 bronzes)

Ba .3

Be .001

Ca .001

Ga .1 (only concentration available on standard;

searched for in 35 bronzes)

Hg .03 (searched for in 35 bronzes)

Mo .03

Nb .003

P .3 (searched for in 35 bronzes)

Sr .03

Ti .001

V .001

Zr .001

TABLES II -VComposition in Relation to Chronological Sequence

TABLE II

Major elements in 97 vessels

Period Cu%

Sn%

Pb%

Shang Hi 87.2 Hiiii 1Qly.yj Hi

(30 vessels) Av 77.2 Av 13.5 Av 7.2

Lo 64.0 Lo 5.9 Lo 0.4

Late Shang-early Chou Hi 84.1 Hi111 Hi111 14 7it. /

(26 vessels) Av 78.1 Av 14.9 Av 4.2

Lo 67.4 Lo 10.0 Lo 0.1

Middle Chou Hi 87.1 Hii i 1 5 81 J . o Hi1 1

1

(16 vessels) Av 74.6 Av 9.7 Av 12.0

Lo 64.7 Lo 5.1 Lo 1.4

Hi 79.7 Hi1 I 1 17 1 Hi1 1

1

(8 vessels) Av 71.5 Av 11.8 Av 14.2

Lo 63.7 Lo 7.9 Lo 0.7

Late Chou Hi 93.3 Hi1 1

1

Hini 1 S f\

v / vessels) Av 78.1 Av/\v 7 Q Av lU.o

Lo 67.8 Lo 1.7 Lo 2.3

Han Hi 87.1 Hi 10.4 Hi 20.0

(4 vessels) Av 79.8 Av 7.6 Av 9.8

Lo 67.4 Lo 4.1 Lo 0.1

Problematic Hi 82.3 Hi 17.2 Hi 21.9

(6 vessels) Av 75.2 Av 12.1 Av 10.0

Lo 67.8 Lo 7.1 Lo 0.2

*Data on vessels of uncertain or recent date (marked in Table A with asterisk) are not included.

54

PQ £

bQ

(73 o

i—

>

bfl

0>

<L)-I—"

173

Q

< o

bO

'a!o3 o"-

<D -3

Q

-I—

I

i73

Q

< •

min

3O43U

13

O -h

o nopNO

oo

O

3I o<u

bO£3(73

GO

bl) 3

j2 43

GO

—1

bOc03

<U GO

bO

03

43GO

o o m ooNO fN m

r-' ON O noin no

q NO pON

bO ^

^432^ lo

-H 72 <U

bt)

Cc3

GO

bOCa43GO

NO

o

U

NO

o oo in ocn NO

O ON NONO >n

On *-<

NOOO OO OO

bo bO bOd fi S303 c3 03

43 43 43GO GO GO

NO

o43u

O ooNOt-~ r-

l

bo s

^43 I O43 r > 1) jh

00 in ONO XT c

;

NO O ON NO* r-'xl- NO in

55

TABLE IV

Types ku and kuang

ku kuang

FGANo.

Date DecreasingCu %

FGANo.

Date DecreasingCu%

07.34 Shang 76.7

17.202 Late Shang-early Chou 76.7

40.3 Shang 75.2

43.9 Shang 73.8

51.18 Shang 72.4

11.51 Shang 64.0

49.10 Late Shang-early Chou 78.0

42.14 Shang 74.4

61.33 Late Shang-early Chou 73.4

39.53 Shang 70.1

38.5 Shang 69.1

TABLE VVessels called ming-cH/-Middle Chou period

Major elements

FGANos.

11.35 \

11.46

11.47

11.58

11.59

11.60

16.248

24.11

Cu%

Hi 87.1

Av 74.8

Lo 65.8

Sn

%

Hi 12.4

Av 9.0

Lo 5.3

Pb%

Hi 22.0

Av 12.5

Lo 5.4

56

CHAPTER IV

FABRICATION

Having reviewed briefly the archaeological evidence and the literary

background, both ancient and modern, relating to the history of metals and

to bronze fabrication in pre-Han times in China, we can now turn to the

examples of the Chinese founders' art in the Freer collection and let them

speak for themselves.

Following chemical analysis, we examined each of the 10 dozen vessels to

get to know its condition, how it was made, how parts were joined, and

everything possible that could be learned about its physical make-up.

Details on individual vessels are given under Technical Observations in the

Catalogue (Vol. I). Many aspects of fabrication revealed there apparently

have not been mentioned in earlier discourses on the subject. In many of

the learned works and catalogues that deal with technical aspects of Chinese

bronzes, it seems that some of the most obvious and most significant

details of design and structure have been ignored. Many of the facts and

observations challenge existing theories and point to the need for new ones.

The following are some of the salient features.

1. Fabrication by casting. All the bronze vessels described in this series

were formed by pouring molten metal into a prepared mold. The very

shapes, the thickness of the vessel walls, the presence of mold marks, and the

internal physical structure of the metal, all testify to a founding (pouring)

technique. The presence of irregular ridges on the bottom of some vessels

caused by the flow of metal into flaws and cracks in the mold reflect the

mobility of the liquid melt. A few of the vessels have vestiges of sprues,

those protruberances formed by the hardening of the metal in the temporary

tubes (ingates) made to conduct the pour from the funnel into the mold

space. We shall see further abundant evidence that the molten metal was

poured into a hollow mold.

2. Surface detail. One of the most striking features of ancient Chinese

bronzes is the richness and sharpness of the relief decor; the technical skill

with which it was executed is most impressive. It seems obvious that over

the course of centuries there was a continuous evolution in the techniques

of mold making just as there was in the forms and decoration of the vessels

themselves.

Characteristically, the decoration of Shang and early Chou vessels was

57

divided into segments by vertical flanges. In early Shang these tended to be

low and thin; and it is quite apparent that they reflect the joins between the

sections of the molds in which the vessels were cast.114 If molds were made by

the application of soft clay to the surface of a model on which the relief

design was carved or otherwise prepared in positive, then the size or width

of each segment depended upon the depth of the relief and on the curvature

of the model surface. Only a small segment of a mold could be withdrawn

intact from a model with deep relief and a tight curve of the surface. Thedecor of early vessels is nearly flush with the surface as if reproducing in

bronze a technique developed in the engraving of softer materials such as

wood, clay or low-fired ceramic. No doubt the evolution and elaboration of

relief decor indicates a growing realization of the potentialities of the bronze

medium and a gradual break with the traditional forms and techniques taken

over from other materials. By middle An-yang times most of the formerly

blank surfaces were being solidly covered with designs; and in later Shang

the ground itself was often completely filled with lei-wen. Major design

elements tended to be arranged symmetrically around dividing flanges or

mold joins. At the same time, the lei-wen of the ground filling were not

always symmetrically disposed as may be observed on the /?w 48. 1 {fig. 20)

(see Vol. 1, No. 5); and close examination reveals that the lines of the lei-wen

were probably made in another operation quite separate from that which

produced the main parts of the design. Evidently, the lei-wen lines were

engraved directly in the mold rather than in the original model; and this

method created a design in two levels. The higher level carries the main

motifs in bold relief while the lower carries all the lei-wen. This contrast is

well illustrated on the ku 43.9 and the kuei 41.S (fig. 21) (see Vol. I, Nos. 9

and 61). The great depth and precision of the lei-wen on a vessel like tsun

51.19 (see Vol. I, No. 16) produces a surface effect like moire when seen in

certain lights (fig. 22).

Within about a century after the establishment of the Chou dynasty,

bronze design began to undergo a gradual stylistic change which in part

seems to reflect a decline in technical ability. In the long run boldness of

design became more important than precision of detail; the technique of

producing fine lei-wen disappeared; and where lei-wen decor persisted, the

lines became coarse and uneven in height and width. By middle Chou they

disappeared completely, the strict compartmentation of the decor began to

break up, and finally the design became more or less continuous around the

vessel. This feature may be seen in such later Chou vessels as the chien 39.5

and the hu 57.22 (see Vol. I, Nos. 94 and 97), decorated in what our Swedish

114 See footnote 134.

58

Hu 48.1. Although the major elements

^**^jfx [.&^- ^cX C* of the design of the tao-t'ieh are symmet-

rical and regular, the fine detail, espec-

ially the lei-wen, is quite irregular andasymmetrical. The lei-wen seems to be a

secondary design employed only to fill

space.

Figure 20

59

colleagues call the "Huai style." On both these vessels the patterns have

some peculiar characteristics such as the joins or stitches in the repeat

patterns which are both uneven and asymmetrical (figs. 23-25). Theyresemble the fragmentary piecing together of odd bits of decalcomania, or

transfer designs on modern chinaware. This is especially true on the braided

rope patterns with fine curved strands which encircle the two vessels just

mentioned. The joins in the braid are often quite unevenly spaced ranging

from 5 cm. to 19 cm. apart. Some of these joins suggest that the design wasimpressed in the mold with a stamp; such a stamp with a positive design

was found among the mold fragments at Hou-ma along with a negative

mold.115 The stamps themselves may have been made of baked earth. Anearthen fao-fieh in the Freer collection has a handle which would make it

easy to use in this way (fig. 27). Another feature of the "Huai style" is the

fine detail that enriches the surface of the entwined animals and of en-

circling braid patterns. On the former are areas of fine raised dots (some-

times called granulation) that look like reptile skin ( hu 51.22, fig. 25); the

fine strand lines of braid patterns look engraved (chien 39.5, fig. 26) but

are really cast.

Another noteworthy feature of the surface detail is the undercutting or

overhang often seen in areas of decor such as those around the oval pro-

truding eyes of the fao-fieh masks, the smaller eyes of birds, and the decora-

tive bosses on geometrical patterns. In some Shang bronzes even the sides

of the fossae in the sunken lei-wen are slightly undercut (fig. 28). This under-

cutting of narrow lines of decor may also be seen in middle Chou vessels

(fig. 29), and it is common in the inscription characters of the same period.

By the end of Chou clear evidence for the use of piece molds becomes less

and less; and by Han times there was evidently a completely new method

of casting bronzes. On four Hen in our collections (see Vol. I, Nos. 1 19-122),

there seem to be no mold marks or pattern joins at all. Although the animals

of Number 121 are cast in fairly high relief like the dragons on earlier

bronzes, the details of hair, eyes, etc., and the diamond pattern on the

border are chased or engraved into the surface. The other three are all quite

plain without cast decor (except for the escutcheon for the ring handle on

No. 122); and again the detail is engraved. The inlaying gold and silver,

which had become fashionable in late Chou times, continued to be popular

in Han.

3. Mold join marks. Many of the vessels show on their outside surfaces,

to a greater or lesser extent, evenly and symmetrically spaced narrow ridges

of metal, mostly vertical in orientation. These are characteristic of any

115Chang Wan-chung, "Hou-ma . . . ," p. 6; Archaeology in New China, Plate LII.

60

Figure 23

Chien 39.5. The faint vertical line (arrows) on a decor area

appears to be a pattern join in a repeat design that might

have been made with a stamp or a repeat device.

Figure 25

Hu 57.22. Two joins, near together, are visible in the

braid pattern which surrounds the vessel. The joins can

be traced upwards (arrows) into the hook and volute

pattern.

Figure 24

Chien 39.5. A mold mark extends downward from the

handle through various encircling bands of sunken decor.

Figure 26

Chien 39.5. Detail ( x 5) of the braid pattern with engraved-

like lines.

61

Figure 27

Baked earth t'ao-t'ieh mask (18.144). The shape of the stem projecting from the back of the mask suggests that

it is a stamp for making bronze molds. Height, 10.5 cm. ; width, 16 cm.

Figure 28

Chia 23.1. The detail (x4) shows not only the crispness of the

lei-wen decor, but also the degree of undercutting.

Figure 29

P'an SC537. A detail from the handle shows

a type of sunken decor often found in

middle Chou dynasty bronze vessels. Thedecor lines are deep and narrow and are

often slightly undercut.

62

Figure 31

Ting 46.31. A mold mark runs vertically through a

cicada motif.

Figure 30

Fang-i 30.54. Mold marks show prominently on the stem

of the handle and on decorative flanges of the lid.

Figure 32

Ting 47.20. Two mold marks radiate from concentric ring

mold marks on the underside of the vessel. The third

radiating mold mark is lost in a repair area.

Figure 33

Ting 47.20. There are two vertical mold marks, one oneach side of the leg.

63

object cast in a piece mold and are caused by molten metal running into the

narrow crevices marking the joins in the mold or in the build-up of the

original model. The prominence of mold marks varies with the degree of

finish and sometimes with the amount of wear on the piece. They are promi-

nent on thefang-i 30.54, especially on the flanges and the capped post of the

lid {fig. 30). As previously mentioned (see Vol. I, No. 38), Noel Barnard

uses the sharp mold marks on this vessel to illustrate his theory of "sec-

tionalism" in relation to early Chinese methods of direct casting.116 On the

thick flanges of vessels of this type the mold marks show as little offsets or

breaks in register at the mold divisions. Within the hooks of flanges,

especially in the larger hooks that hang from kuei handles, webs of metal

that partially close the hooks indicate the location of mold joins. Poor

register of mold sections causes the peculiar vertical ridge down the middle11GBronze casting . .

. ,Chapter 5.

64

of a cicada in the decor of ting 46.31 {fig. 31). There are many small

aberrations of this kind along mold joins. On larger ting of the late Chouperiod like 47.20 {fig. 32), it is not uncommon to find a circular ridge on

the bottom with three equidistant and radiating mold marks. Legs of ting

quite often have vertical mold joins on either side (fig. 33). There seem to be

mold joins of different types. Those on the vessels of Shang times definitely

follow the flanges of square and round vessels. A slight vertical depression

on the stepped-back neck ofyu 40. 1 1 seems to be a continuation of the mold

join that runs along the edge of the flange {fig. 34). On vessels of animal

form the prominent mold marks coincide with the median longitudinal

plane.117

As mentioned above, mold-join marks are barely visible on highly

finished bronzes, but they stand out on vessels of poorer quality - especially

on that class of objects called "ming-chi" which were intended purely for

funerary and burial purposes. These latter are important, however, because

they give clues to the method of fabrication of better vessels. A fine example

is the lead chiieh mentioned in Chapter II. Apparently after casting no

attempt was made to obliterate the four mold marks which run vertically

on the sides of the vessel at the quadrants {figs. 35-38). The principal joins

(true joins) are in the plane bisecting the pouring spout and the point of

the rim (long axis), and they run downward along the sides of two of the

legs. Secondary join lines (pre-assembly joins) show along the handle and

the flange opposite. In addition, there is evidence {fig. 38) of a horizontal

mold mark where the body and legs join. 118 On some of the Freer chiieh there

are traces of marks which indicate similar horizontal mold joins. Among 24

chiieh examined by the writer at the Royal Ontario Museum, however, none

showed vestiges of horizontal mold marks where the legs join the body.

More localized mold marks are associated with handles, legs and appen-

dages. For example, on kuei with integrally cast handles, there are sometimes

vertical parallel ridges on the vessel sides which connect the outsides of

the upper and lower joins of handle to body, as is shown so well on kuei

60.19 {fig. 39). These lines must relate to the special mold core for the

channeled handle in which residues of the core are often found still in place.

117Noel Barnard calls such mold joins the "true joins" because they "are the lines of juncture formed be-

tween adjacent mold pieces - each such mold-piece being a separate unit." He also speaks of "pre-assemblyjoins" which "are the lines of juncture formed by sub-sections of a mold piece which have been cementedtogether by clay slip. Such sub-sections totaling two or three adjacent pieces, were each separately impressedfrom the clay model presumably because of difficulties in withdrawing a larger segment (from the model) as asingle unit" (private communication).118 Noel Barnard, who has made a special study of this lead chiieh (private communication) thinks this piece

tells a lot about mold assemblies for vessels of this type. He believes the body was cast in a two-piece (four-

division) mold assembly but that the legs were simultaneously cast in a seven-piece mold, and the horizontalline indicates the location of the join between vessel and leg mold assemblies. He cites a group of fragmentsof a piece mold of a chiieh excavated at An-yang pictured in Archaeology in New China (Plate XXXI), as further

evidence of horizontal divisions in chiieh piece molds.

65

Figure 37

Figures 35-38. Lead chiieh SC546. Figure 35. This early

Western Chou dynasty vessel shows prominent moldmarks at the quadrants. The mark that starts under the

pouring spout can be followed along the edge of a leg to the

tip. Figure 36. On the opposite side of the vessel the moldjoin marks can be traced equally well from the edge of the

pointed rim to the leg tip. Another mold mark also shows on

Figure 38

the back edge of the handle. Figure 37. A mold join that

shows prominently on the flange opposite the handle con-

tinues upwards to the rim. Figure 38. On the underside a

circular horizontal mold join connects the tops of the legs.

This marks, apparently, the separation between the moldassemblies for the bowl and that of the legs.

Figure 39

Kuei 60.19. Two vertical mold marks on the side of the vessel seemto be continuations of join marks that rim the inside of the handle.

Such mold join marks are commonly found on vessels of this type.

Even thin-wall vessels of the late Chou and Han period like the covered

food vessels of the type fu which are made in intricate shapes with muchoverhang show vestiges of mold join marks along their hipped corners.

By Han times mold marks seem to disappear from ceremonial vessels at

the same time that we notice changes in decor and other characteristics of

surface detail. As mentioned above this may reflect a change-over from

piece mold casting to an entirely new method of vessel production. It is

possible that this marks the advent of cire perdue casting in China although

direct evidence for that technique is lacking. There is no evidence of multiple

pouring gates and vents nor of pin-type chaplets which characterize cire

perdue foundry practice.

4. Sprues. A sprue is the foundryman's term for the pipe or duct provided

in each mold assembly to conduct molten metal from the funnel-like open-

67

ing or pouring cup in the top into the mold cavity. 119 On cooling, the metal

forms a stem which is cut off in the finishing operation. Most of the vessels

are so highly finished that no trace of a sprue can be found. The legs,

especially those of the type kuei and ting, often show on their bottoms

evidence that the vessels were cast upside down and that the legs served in

part as sprues at the time of fabrication. The bottom of the legs of huo 1 1 .630

{fig. 40) are stepped up on one side showing that the higher portion is in each

case the remains of a sprue. The legs of kuei 60.19 (fig. 41) are stepped in

the same way ; but, in addition, the higher side has a shallow depression which

indicates the metal was poured just to this level and the depression wascaused by shrinkage of the melt on cooling. Some of the Han vessels of the

type hu have a short ridge on the underside, and the top of the ridge is

rough and irregular which suggests that it is a broken-off sprue. Such

ridges on the underside of hu 15.103( //g. 42), hu 61.32, and p'an 11.35

{fig. 43) are good examples; one lien 51.5 and one kuei 11.58 [fig. 256,

Chapter VIII) also show this feature. On many vessels previously described

the sprue stump seems to be in the line of a mold join, but on these flat

bottom vessels with centered ridge there is no evidence of a mold join. 120

The ridge on the underside of one of our hu 66.14, shows definite marks of

the chisel used to break it off ( fig. 44). On vessel bottoms where the ridges

do not show, there was no need to eradicate them.

The size (cross-section) of the sprue no doubt varied with the amountof metal to be poured; a narrow sprue would not allow sufficient molten

metal to pass through to fill a large casting volume at the necessary rate of

flow.

Stumps of sprues are often seen on patches or repairs made soon after

fabrication. Apparently, when a hole in a vessel was caused by failure of

molten metal to flow in completely, a crude repair was made by building a

special small mold around the hole and pouring in a patch. In some of

the more crudely made patches the stump of the sprue plainly shows; for

example, on the underside of p'an 11.35 {fig. 43) and hu 24.12 (fig. 142)

(vide infra).

5. Shrinkage depressions. Small depressions here and there on vessels in-

dicate that the Chinese foundrymen had to cope with shrinkage, as have

foundrymen of all cultures since the beginning of metallurgy. Most materi-

als, including metals, contract on cooling at the point of change from the

119In modern foundry parlance any vertical conduit in a mold leading from the "pouring basin" or "pouringcup" into the mold cavity is called a "down sprue"; any horizontal conduit is a "runner" and the point ofjunction of a runner with the mold cavity is called the "ingate". For other terms used in casting see HenryHodges, Artifacts, pp. 69-73; also Herbert Maryon, Metalwork and Enamelling (1954), pp. 212-228.120On the underside of a round hu in the Royal Ontario Museum (926.21.231) there is a wide wedge-shapedsprue tapered downward which indicates that it was the lower part of the pouring cup.

68

Figure 40

Huo 11.630. The underside of the single

hind leg of the animal-form vessel has

a low shelf or raised area which could be

taken as evidence that the leg served as a

sprue for the molten metal during

casting.

Figure 43

P'an 11.35. The low ridge at the center of the

underside appears to be a broken-off sprue

indicating the placement of the pouring gate

at the time of casting. The short sprue in the

repair area, right, indicates that repair metal

was also poured into position.

Figure 41

Kuei 60.19. The depressions in the underside of the legs indicate that

they served as sprues.

Figure 42

Pien-hu 15.103. A ridge in the center maybe the stump of the sprue.

Figure 44

Hu 66.14. The ridge on the bottom, perhaps the stump of a sprue,

bears marks which suggest it was cut off with a chisel-like instrument.

69

Figure 45 Figure 46

Kuang 61.33. The depressions on the inside of the cover

opposite the protuberances on the upper side served to

keep the vessel walls thin and hence to minimize shrink-

age during the casting operation.

Huo 42.1. Hollows on the inside of the lid

correspond to the high relief facial features

on the outside. This thinning of the vessel

walls minimized shrinkage at the time of

casting.

liquid to the solid state. Shrinkage occurs everywhere, but, due to gravity,

manifests itself mostly in the upper parts of the casting so that the location

of the shrinkage depressions sometimes indicates the direction of pour.

Depressions at the base of the capped posts of some of the chiieh seem to

support the theory that they were cast in the upright position and that the

stems served as sprues. There are shrinkage depressions on the bottoms of

some legs and under some knob handles of covers, which indicates they were

poured up-side-down. That the early Chinese bronze casters were aware of

shrinkage and knew how to cope with it is shown by their practice of

shaping the core to leave depressions in the vessel wall directly behind

prominent protuberances in the decor. Examples can be seen in the inside

cover of kuang 61.33 (fig. 45) and inside the cover of huo 42.1 (fig. 46). In

addition, the common use of clay cores in legs and handles was probably

not only to conserve precious metal but also to keep the volume of metal

small and thus reduce shrinkage.

6. Criss-cross lines and miscellaneous markings. Some vessels have

peculiar ridges, markings, openings and appendages which apparently have

nothing to do with decor or function but relate to devices used in fabrica-

tion. The most difficult to explain are the low ridges that cross and often

criss-cross the underside of the bottoms of some of the vessels. Sometimes

these ridges give the distinct impression that they are scribe marks casually

made in the clay core prior to casting. They remind one of the vent holes

70

the housewife jabs in the top crust of pastry to let out steam. They makedifferent patterns of which the most common is a square mesh design like

the one seen on the underside of kuei 1 1 .38 (fig. 47). The ridges are unevenly

spaced; and some lines going in one direction seem to cross over lines going

at right angles. From mesh to mesh the lines are uneven in height and thick-

ness. It is perfectly plain in this case that the ridges were caused by scratch

lines in the clay core which had not been completely removed from the well

of the foot. On the underside of kuei 60.19 (fig. 48) the scribe marks form

a peculiar geometric pattern of nested right angles; and the underside of

pL

an 11.35 (fig. 43) is interesting because the criss-cross lines are not at

right angles but at acute angles, and moreover are crossed by a cut-off

sprue. It is possible that these scratch marks were made for the purpose of

providing better escape of gases at the time of casting. It has also been

suggested that they were employed to prevent slippage to or insure proper

register of the mold sections.

O. Karlbeck (private communication) has put forward the suggestion that

the criss-cross lines are scorings in the clay core made by threads or strings

that were used to suspend the foot core in the mold to permit molten bronze

to flow between upper and lower cores and thus form the bottom of the

vessel. This presupposes that the bronze was cast upside down. Karlbeck

points out that on some of the bronzes which bear criss-cross patterns there

is evidence of cross-over of strings going at right angles, and occasionally

there are signs of tie-knots in the strings. This theory partially fits the facts

for vessels with flat or slightly curved buttoms, but it fails to explain the

mesh marks on other vessels. For example, on a bronze chia in the City Art

Museum, St. Louis (C.A.M. 221.50), criss-cross lines intersect on the curved

underside of the bulbous legs (fig. 49). On another li-ting in the Ashmolean

Museum (No. 1956.861) similar encircling lines on the bulbous legs are

interrupted by three symmetrically placed chaplets (fig. 50). On the bottom

of our kuei 11.38, the criss-cross lines are too uneven and irregular to be

caused by taut string lines. To further complicate the problem, the writer

has seen in a private collection a matched pair of Chou dynasty kuei, one

with criss-cross lines on the bottom, the other with plain bottom.

On the bottoms of some vessels the corners of the legs are connected by

raised bands or ridges which have a strap-like appearance; Q.g.fang-i 30.54

(fig. 51), fang-ting 50.7 (fig. 52) and ting 46.4 (fig. 53). In each case they

seem not to serve as a decorative feature but to reflect some aspect of fabri-

cation or function. There is one vessel, kuei 15.102 (fig. 54), where the low

relief design underneath is not just criss-cross but represents a coiled dragon.

Here the technique and perhaps purpose is similar, but done with more

71

Figure 47 Figure 48Kuei 11.38. Detail of underside showing criss-cross Kuei 60.19. Detail of the underside showing criss-cross

ridges or "scribe lines" partially hidden by remains of the lines which form a pattern of nested right angles,

original earthen core.

Figure 49

Chia, City Art Museum, St. Louis, 221.50. The underside of

the bulbous legs of an early Chou bronze are crossed with lowridge lines that intersect on the leg surfaces. (Called li in St.

Louis catalogue. Courtesy City Art Museum of St. Louis.

Gift of J. Lionberger Davis.)

Figure 50

Li-ting, Ashmolean Museum, Oxford,

1956.861. The ridge lines partially encircle

each bulbous leg. Several chaplets are also

plainly visible. (Courtesy, Department of

Eastern Art, Ashmolean Museum.)

72

Figure 51 Figure 52Fang-i 30.54. A detail of the underside shows criss-cross Fang.ting 50 . 7 . Criss-cross ribs on the underside con-ribs or thicken.ngs which seem to have some functional

necting the ,egs are faMy common on vessels of thispurpose. type

Figure 53

Ting 46.4. The triangular ribbing on the underside is

slightly offset to the legs.

Figure 54

Kuei 15.102. On the underside the raised lines do not form

the usual criss-cross pattern but represent a coiled

dragon.

Figure 55

Kuei 11.49. On this archaistic vessel the criss-cross lines

are even and regularly spaced indicating that they are

imitative rather than functional.

imagination. 121 On archaistic bronzes made in the Sung and later periods,

the casual criss-cross lines are formalized into regular and evenly-spaced

mesh and diamond-shaped patterns and become part of the decor. Theyappear to be imitative rather than functional. A good example is on the

bottom of kuei 1 1 .49 {fig. 55), which is judged to be recent.

7. Buttresses. On some of the larger vessels with high-set bottoms, wedge-

shaped reinforcements or buttresses are placed at the angle between the foot

and the underside of the bottom. As in tsun 55.1 {fig. 56), the buttresses maybe unevenly spaced. In kuei 60.19 {fig. 48), they are a continuation of the

legs. Long thin buttresses are occasionally found on the inside of the foot of

ku which have small but high-set bottoms. These do not seem to be intended

for strengthening purposes, but may be related to some practical aspect of

mold fabrication. Professor Douglas C. Williams of the Ohio State Univer-

sity (private communication) has suggested that these thickenings may be

analogous to "cracking strips11

used by modern foundrymen to prevent

cracking or "hot tearing" at sharp angles during cooling of the metal.

8. Cross-like perforations in ku. Various kinds of openings, usually

squarish in shape, are sometimes located just below the bottoms of the

vessels at the juncture of foot and bottom; a good example is tsun 51.19

{fig. 57). They are sometimes found at the base of flanges that serve for

handles of covers. Close inspection of these perforations shows that they

were probably formed by protruding members of mold cores which served

to keep the mold and core separated at proper distance.

Especially interesting are the perforations in the shape of a Greek cross

that are found on many ku. The bottom of a ku is always set high within the

narrow central section just above the plain circular band at the top of the

foot. The perforate crosses are invariably set just below the bottom on

opposite sides of the vessel {fig. 58). Sometimes they are plainly visible

although filled in partially or solidly with metal {fig. 59). O. Karlbeck says

(private communication, 1960) he has seen two or three ku with four

crosses; others with three. He has also seen ku with vertical slits alternating

with the crosses. One of our ku (11.51) has no crosses, and there are two

without crosses in the Museum of Far Eastern Antiquities in Stockholm

(Nos. K 12974 and K 14803). Of seventeen ku seen in the Royal Ontario

Museum nine have crosses; eight have none.

There has been some speculation about the purpose of these crosses.

They can hardly be decorative. Noel Barnard (private communication)121A kuei with similar design belongs in the Pillsbury Collection (Karlgren, Pillsbury Collection, Cat., No. 34,

plate 51) ; there are others of this kind ; see Vol. I, pi. 70. There are other unusual features. On the underside ofan An-yang kuei recently described by Ch'i T'ai-ting in Kaogu 1964, No. 1 1, p. 592, fig. 1, No. 5 and PI. XII,

No. 1, are two pairs of irregularly spaced criss-cross lines forming 9 squares; located in the center square is

a small human mask cast in low relief.

74

Figure 56

Tsun 55.1. Triangular thickenings or buttresses at junc-

ture of bottom and foot were employed perhaps to

counteract "hot cracking" at the sharp angle where

bottom and foot meet.

Figure 57

Tsun 51.19. Three slot-like perforations are located

symmetrically where the foot of the vessel joins the

bottom. These may be points where spacers of some kind

were used to keep the mold core in place. One of the

irregular thickenings in the wall of the foot shows (left ).

Figure 58

Ku 40.3. A perforate cross is commonly seen in the plain

band that marks the waist of vessels of this kind. These

crosses may be caused by some device employed to keep

mold sections in register.

Figure 59

Ku 51.18. Crosses on opposite sides of the waist are

closed with thin webs of metal flush with the inside sur-

face. They may be formed by spacers which extend fromthe outer mold.

75

believes that the cruciform perforations are the imprint of what he calls

"core extension spacers,'1

devices to keep the upper and lower core of the

mold in proper alignment. O. Karlbeck concurs, "In my opinion the

flanges using the cross-shaped perforations had a double function. Thevertical arms kept the core in position and by means of the horizontal arms

the craftsman could detect whether the two adjacent mold sections had been

rightly assembled. All he would have to do would be to measure the dis-

tance from the top of one horizontal arm to the top of the decorated part of

the mold section and compare it with a measure taken from the other arm"(private communication, 1960). Wan Chia-Pao has expressed the view that

the cross-shaped perforations were caused by "bronze chaplets of cross or

rectangular shape, equal in thickness to the wall of the ku, which were in-

serted between the external moulds and the mould of the ring foot core.

These chaplets would normally fuse with the poured metal in the casting and

produce an unbroken surface. But when fusion was incomplete, corrosion

could easily occur and the chaplet metal either partially or totally dis-

appear." 122 The author, however, has examined dozens of ku in all conditions

of preservation and found no evidence that would support Wan Chia-Pao's

assertion. James Cahill (private communication) points out, however, that

cross-shaped openings appear also in pottery vessels casting some doubt

that they are to be explained purely in terms of bronze casting technique.123

To confound the issue, Watson 124 shows an early Shang dynasty ivory

goblet which was excavated in 1952 at Erh-li-kang near Cheng Chou, which

is of special interest because it has a cross-shaped perforation on the foot

very much like the crosses that appear on the bronze ku of later date des-

scribed above. These crosses are truly enigmatic.

9. Lock-on casting of legs. Some vessels, especially those of the types

kuei, fu and /' have stubby solid legs which are cast as part of the vessel.

Other vessel types like the ting, chia, chiieh, and huo have three or four legs

which make up nearly half the vessel height. In three and four-legged

vessels of earlier times, the legs are solid cast as an integral part of the

vessel. Also some early chia which have clay-cored capped posts and legs

are apparently cast in one piece. Chia 23.1 is one good example. The legs of

some of the late Chou ting, however, especially the larger ones are often cast

separately and joined to the vessel by methods which will now be described.

The legs, and even handles, of some of the vessels are not solid metal but

are thin walled and filled with a core of partially baked earth or clay. In

122 English summary, Part I "The process of casting the Ku-beaker," p. 120.123A Shang pottery vessel with perforate cross found at Cheng-chou is illustrated by Cheng Te-k'un in ShangChina, Plate 29a.

^^Archaeology in China, Plate 38.

76

some legs and handles the clay core is partially exposed; but on others it is

surrounded by metal and so completely concealed that the core can only be

detected by drilling into the interior or indirectly by X-radiography.

The presence of lock-on members is revealed only by a single narrow

seam at the join where it is often well concealed by corrosion and dirt. Aparticularly good example is the largest tripod vessel or ting 47.20 in our

collection. The three bulbous legs of this cauldron are filled with reddish

terra cotta; each leg bears two vertical mold marks, one on either side {fig.

33). Careful cleaning of the point of juncture between one of the legs and

the body revealed not only evidence of a seam but also the existence of an

irregular line of flash or spill-over of metal from the body on to the leg

{fig. 60). The thin-walled and clay-cored handles of this vessel are joined

in the same way {fig. 61). There is no sign of solder at any of the joins. Onan exhibition piece like this which cannot be deeply scraped and probed,

it is difficult to tell just how the joining was done.

A partial answer to this problem was furnished by a similar piece in the

study collection (SC587) which was purchased in China, and brought to

America many years ago; two of the legs had broken of! where they

joined the body {fig. 62). Examination showed that the bottom of the ting

is a modern repair made of sheet copper; the legs were simply joined to it

with soft tin-lead solder and the whole reconstruction was concealed with

paint and false patina. The two legs had broken off because of failure of the

solder to stick to the corroded original parts, but the third leg fortunately

held in place. There is every reason to believe that the rim and handles are

parts of an authentic bronze and that they belong together. Further examina-

tion of the two detached legs revealed that the top of each leg is enclosed by

a portion of the side wall of the original vessel and that there is a thin but

distinct seam between vessel wall and top of leg, just like that seen on our

ting 47.20 above. Short dribbles and runs of metal from the vessel wall over-

flow the seam. Probably the original edges of the side walls that had

broken away along with the leg were rough and irregular, but these were

trimmed and smoothed to facilitate the joining of the legs to the false

bottom.

A final answer to the question of how these legs were attached was

dramatically revealed by slicing one of the legs in half lengthwise with a

circular diamond-filled lapidary saw. The cross-section {figs. 63 and 64)

reveals that the leg is filled with soft baked clay, mostly dark gray in tone but

reddish in certain places. The top of the leg is enclosed by a section of the

vessel wall which is keyed or locked into the top edge by a cast-in collar.

These observations raise new questions. How was the casting-on accom-

77

Figure 60 Figure 61

Figures 60 and 61. Ting 47.20. A detail of the join of leg likewise shows a thin line or spill-over of body metal onto body shows spill-over of metal from the body on to the to the handle,

leg. Figure 61. A detail of the join of handle to body

Figure 62

Figures 62-66. Ting SC587. This vessel, formerly privately

owned, lost two of its legs in shipment. Examination

showed that the three legs and the upper portion are

genuine, but the entire vessel bottom, almost to the

decorative circle, is a modern replacement made from

sheet copper. The tops of the legs, however, were covered

by a portion of the original vessel wall. Failure of the new

Figure 63

solder joins to hold to old metal caused the assembly to

break apart. The original parts have been invaluable for

study purposes. Figure 63. The sections of one of the legs

after cutting in two show clearly that the vessel memberwas cast to the leg in such a way that the two interlocked

and held fast. The leg has a core of semibaked clay which

is dark gray in tone. The thinness of the metal walls and

78

Figure 65

the partial modeling done in the core should be noted.

Figure 64. A further detail of one leg section, in addition to

showing the casting-on method of joining body to leg,

shows small voids in the clay core; it also shows numerous

scattered inclusions of lead in the bronze matrix. About x 2.

Figure 65. A radiograph of the sectioned leg of the ting

and an unsectioned leg (center) shows the even thickness

of the metal and the thin (dark) areas where core supports

("supporting nails" or "core extension spacers") protrude

into the metal. Figure 66. A metallograph (x 100, bright

field, unetched) made from a sample cut from the area of

the join shows cuprite in the line of join between bodymetal (upper) and metal of the leg (lower). Although the

body metal was molten when it came in contact with the

precast leg, there is no sign of fusion of the two members.

79

plished? How were the molds constructed? The obvious answer is that the

legs were all pre-cast with clay cores but with tops left open. The core of

each leg was then channeled inside the edges of the tops of the legs, andfinally the legs were set symmetrically into the sides of a mold that had been

prepared for the vessel. The fluid metal flowed into the leg channels and,

when set, locked them securely to the vessel. The mold for a single ting leg

of this kind coming from the Hou-ma foundry site is pictured by ChangWan-chung.125 Some "supporting nails" or core-extension spacers, as

Barnard calls them, can be seen clearly both in Chang's drawing and on the

left in our figure 63. While the metal was being poured, the inner and outer

molds for the leg were held apart by means of these "supporting nails" and

by a chaplet at the bottom rear of the leg. The other "supporting nails" can

be clearly seen as dark rectangular patches on a radiograph of the leg

sections (fig. 65). Chang also shows a drawing of the outer mold for a ting in

which a hole has been left for insertion of a handle to be cast on to the vessel.126

Chemical analyses made of the leg metal and of the body metal are shownin Table VI. They are not from the same melt because of the time interval

between casting the leg and fixing it in the mold for the casting of the vessel.

TABLE VI

Analysis of separate members of ting SC587

Leg BodyCu 67.2% 69.9%

Sn 8.2 10.3

Pb 23.6 18.8

Total 99.0 99.0

A metallograph of a sample from one of the legs (fig. 66) shows no

evidence of brazing metal or of solder at the juncture for the two members;

there is only a thin line of cuprite corrosion product between them. The legs

of another bronze, a late Chou ting in our study collection, shows the same

method ofjoining. This ting (SC613) is crudely fashioned with legs that have

a peculiar lap join where they are attached to the body. A sector was cut

from the whole vessel in such a way that two of the legs were bisected verti-

cally. This revealed (figs. 67 and 68) an inner but incomplete leg with clay

core locked on in exactly the same way as the legs of the ting (SC587)

125"Hou-ma fig. 1, p. 39 and fig. 6, p. 41.

126"Hou-ma . ..

," fig. 1 1, p. 42.

80

Figure 67

Figures 67 and 68. Ting SC613. This roughly cast vessel is

of special interest because of its locked-on legs. When the

round vessel was sectored, it was found that each of the

original legs was broken off at an early date and later com-

pleted by an overlay casting. Figure 68. A vertical cut

Figure 68

through the leg reveals clearly how the repair leg was cast

over the incomplete original. A two-piece core was used for

the original leg. There is a peculiar void in the bottom of

each of the repair legs, perhaps a casting flaw.

Figure 69 Figure 70

Figures 69 and 70. Separate leg of a bronze ting in the

Royal Ontario Museum (ROM 933.23.64). Figure 69. Theleg (shown about half size) has been wrenched from the

vessel but a portion of the original vessel wall was carried

with it. The thin-walled leg is filled with a core of semi-

baked earth. A small sample to be used for study purposes

was cut from the edge where the leg joins the metal (arrow).

Figure 70. A metallograph ( x 50, bright field, unetched)

shows the line of join between body metal (right) and leg

metal (left). The gray area in the join is cuprite; cuprite also

permeates the metal of both members. (Both photos,

courtesy Royal Ontario Museum, University of Toronto.)

81

described above. But there was another surprising circumstance. Each of

the inner legs was found to be encased in a second leg which apparently

had been cast around the original leg. This explains the peculiar lap join

where the legs join the body; apparently the original legs were faulty or

incomplete after the first casting and the overlay legs were cast on for repair

purposes. It could also be that the original legs were broken off in use and

that this is a much later repair.

The same kind of construction is shown in a single detached leg of a ting

in the Royal Ontario Museum (fig. 69). A portion of the thin vessel wall is

still attached to the leg and the edge of it forms an overhang. In a metallo-

graph (fig. 70) taken by Dr. Ursula Martius of the Ontario Research

Foundation, a clear line of contact separates body and leg. Again there is noevidence of brazing metal or solder. X-ray fluorescence analysis by Dr.

Martius of leg and body parts of this object (Table VII) shows that the

different members have quite different composition.

TABLE VII

Analysis of separate members of ting ROM 933.23.64

Leg Body

Cu 74% 59%Sn 14 26

Pb 1

1

14

Zn 0.5 0.5

Another detached leg of a ting in the same museum (ROM 933.23.63)

(fig. 71) is even more complex in construction. Not only is the leg cast to

the body, but the leg itself is cast in two parts which are mechanically

locked together (figs. 72-74). The horizontal join can easily be seen on the

outside of the leg. As is shown in the cross-section, the lower part of the leg,

which was cast independently of the vessel, is made of nearly pure copper.

The purpose of the upper part of the leg (middle section of the detail) seems

to be to lock the copper leg and the vessel together. It is an unusual con-

struction but not an illogical one. These observations lead us to a better

understanding as to how interlock-casting was done and how the molds

were constructed.127

127The employment of casting-on techniques for bronze fabrication is not unique to early China. As shown byHans Drescher in his recent work Der Vberfangguss, which can best be translated "Overlay Casting," it wasemployed extensively during the Bronze Age in Europe. The technique was used not only for fabricating pur-

poses but also for making repairs to incomplete castings or to castings broken or damaged in use. Many early

swords which were thought to be cast in one piece have been shown by X-radiography to have cast-on handles.

In the Freer Gallery collection there are two Chou dynasty bronze weapons, one originally made with an iron

point (34.11); the other with iron blade (34.10); X-radiography has shown that in each case the handles of

these weapons were cast to preformed iron parts.

82

Figure 71 Figure 73 Figure 74

Figures 71-74. Leg of another large ting in the Royal

Ontario Museum (ROM 933.23.63). Figure 71. A portion

of the sidewall of the original vessel caps the earthen core

of the leg. The leg is cast in two parts, the decorated upper

portion which bears traces of a surface coating of tin, and

the undecorated lower portion. The location of a section

cut out for metallographic purposes is indicated by arrows.

Figure 72. The saw cut shows the lines of join of the three

members, the body metal, the upper leg, and the lower leg.

Figure 73. The drawing of the leg illustrates the following

features: the leg, which is made in two parts, is mechani-

cally joined to the body metal (marked "A") by inter-

locking fillets. The upper leg (marked "B") in turn is

locked to the lower leg (marked "C") by another cast-in

fillet. The upper leg was apparently the first member to be

cast and the vessel and lower leg were cast to it. The twocavities (shown black) were probably originally raked out.

Some brown material (marked "D") which still lines the

interior surface is hard and possibly consists of slag.

(Drawing after a sketch by R. M. Organ.) Figure 74. Thecut-out wedge which was mounted and polished shows the

three sections of metal locked together. The top member is

body metal, the center is the top of the leg, and the bottomis a portion of the lower member. It is obvious from the

light tone of the lower section that it is made from an

alloy that is different in composition from the other

members. Metallograph, not etched, x7. (Courtesy,

Royal Ontario Museum, University of Toronto.)

83

Not all bulbous-top ting legs are cast on. A small Chou dynasty ting

(11.60) has bulbous legs which are cast integrally with the body (fig. 75).

The legs are hollow but are open-faced on the inside, and the hard clay in the

hollow appears to be original core, not earth fill from burial. The construction

is much like that of hollow handles of many of the vessels of the type kuei.

10. Engaged handles. Several of the vessel types, especially the kuei, ehia,

chueh, huo and kuang have loop-shaped handles cast as an integral part of

the vessel. The handles of most kuei are like 20th-century bouillon-cup hand-

les and are placed on opposite sides of the open vessel. Usually they are sur-

mounted by animal heads in full round and in addition bear on their under-

side decorative hooks or lobes. They are almost invariably channeled on the

inside, and the channel is filled with baked clay, apparently part of the

original core. Sometimes, as in kuei 60.19 (fig. 39), there are parallel,

vertical ridges on the side of the bowl connecting upper and lower joins. It

has been suggested that D-shaped clay cores were used with these engaged

kuei handles, and that these have left their marks on the vessel side. Thecores, except that part protected by the channel, were knocked out after the

casting was made. Some thinner handles, without channels, like those onchueh and ehia have vertical mold marks on the inside (fig. 76).

1 1 . Lock-on casting of handles. It has already been mentioned that the

handles of the large ting 47.20, as well as the legs, are clay cored and that the

vessel seems to be cast on to the handles (fig. 61). The same is apparently

true of the handle of the large kuei 38. 20.128 In a related method one or moreholes are provided in the vessel side, and the ends of the handles pass

through these and are locked by bosses on the inside. The rivet-like bosses

which lock the handle of ehia 07.37 (fig. 77) and the lid grip of ehia 35.12

(fig. 78) were apparently secured in a single lock-on casting operation. Onboth vessels the bosses are decorated by cast-in scroll patterns.

12. Brazed-on handles. A third type of handle is cast separately and

joined to the vessel with a filler metal or kind of hard solder made from a

copper-tin-lead alloy. Hard soldered or brazed-on joins are the rarest kind,

having been used on only six vessels in the series, four of them to fix handles.

The handles of the much corroded kuei 24.1 1 is one of these. As shown in

detail photographs (figs. 79 and 80), there is a separate fill or ribbon of

metal at the joins between handle and body. A section cut from the lower

end of a handle showed that it is hollow and contains its original hard clay

core. The cut-off end showed that the filler metal had run into the space

between the handle and the vessel to make a firm bond with the inside

128Unfortunately it has not yet been possible to procure a study piece which could be sectioned to show the

exact method used in interlock casting of handles.

84

Figure 75

Ting, 11.60. The legs on some of the ting were cast

integrally with the vessel. There is no indication of a seamwhere leg meets body.

Figure 77

Chia 07.37. The separately cast handle is locked to the

vessel by permitting the molten metal which formed the

handle to flow through two holes and form rivet-like

bosses on the underside.

Figure 76

Chia, 23.1. A mold mark shows prominently on the inside

of the handle.

Figure 78

Chia 35.12. The bird-form hand grip is secured to the lid

by rivet-like bosses on the underside formed by permit-

ting molten metal to flow through two holes spaced

2.5 cm. apart.

85

surface of the handle (fig. 169). Analysis of the filler metal showed that its

tin content is much higher than that in either the handle or the vessel;

hence it had a much lower melting point. It is not known how the filler

metal was run into the join. There is no evidence that any kind of form, damor mold was used to contain it. Professor Cyril S. Smith has suggested that

the brazing metal may have been wiped or buttered on in a pasty state.129

A similar technique is used to join the handles of the chien 39.5 (fig. 81),

but the shoulder formed by the hard solder is untidy and irregular. Thecleanest join of this kind is on yu 30.26, where hard solder is used to join the

four protruding arms (fig. 82). Here the ribbon of hard solder is straight-

edged, narrow, and beveled at an angle near 45 degrees. It is clearly shownin an X-ray (Ch. VII, fig. 207) that the appendages are fitted to the vessel

over a square boss and the hard solder is used principally to re-enforce and

conceal the join. It was possible to lift away a bit of the hard solder where it

had not adhered well and use it for analysis. The composition is: Cu 71.8% ;

Sn 13.0; Pb 12.6; total 97.4, which strangely is very close to that of the bodyalloy. Other vessels with brazed-on handles are hu 13.21 and kuang 39.53

(fig. 83). The handle of the i 61.31, appears to be fixed by interlock casting,

but the top of the spout is brazed on. Thus, more than one technique of

joining was employed on a single vessel.

H. Maryon makes it clear that separate parts made from copper and its

alloys cannot be joined by simply heating them to incipient fusion (auto-

genous welding) or by hammering, because of the formation of a film of

copper oxide on the metal surface.130 Except for mechanical joining by use

of rivets or by casting-on, copper and its alloys can only be joined by

hard or soft solder which have a melting point lower than the parts to be

joined. Also the contact surfaces must be made scrupulously clean and free

from oxides with acids or fluxes so that the liquid soldering alloy can wet

the surface of the parts to be joined. The ancient Chinese must have used

fluxes of some kind in their brazing, but we have no idea what they were.

Modern brazing alloys consist of copper and zinc. Soft solders are mainly

alloys of lead and tin. Lead-tin solders were known in Roman times and

perhaps even earlier in the Near East, but they do not seem to have been

used by early Chinese metal workers, although the separate metals, lead and

tin, were available to them.

13. Escutcheon-and-ring handles. In vessels of the late Chou and Hanperiods, a quite different type of handle evolved. It is a flat plaque modeled

in low relief in the form of a demon mask with two prominent eyes and a

129A metallographic study of the filler metal made by Professor Smith is summarized in Chapter V.130"Fine metal-work," p. 652.

86

Figure 79 Figure 80

Figures 79 and 80. Kuei 24.11. Figure 79. The handle is brazed to the vessel. The filler metal

plainly shows at the join. (An all-over view of the kuei is shown, Plate 81, Volume I.) Figure 80.

The lower part of the handle has been cut away to reveal the filler metal used to braze the handle

to the vessel. The channel of the handle is filled with mold core (arrows). The vessel is shownupside down.

Figure 81 Figure 83Chien 39.5. The edges of the brazing metal (arrows) at Kuang 39.53. The handle is brazed to the vessel. Thejoins of handle to the body are untidy and irregular. filler metal shows plainly at the lower join.

curved beak of proboscis that forms the loop for a ring. Some of these

escutcheons such as those on tui 24.13, lien 23.2 and lien 51.5 are cast in-

tegrally with the vessel and in each case the ring handles are split indicating

that they were fixed on after casting. The escutcheons on the three hu, 1 5. 103

(fig. 84), 56. 1 5 (fig. 85) and 61 .32 (fig. 86), however, are separately cast and

fixed to the vessels by some method not yet fully understood. The escutcheon

on hu 61.32 is separated enough from the vessel side to permit the insertion

of a wafer-thin razor blade to a distance of about 1 mm. On the inside

opposite the nose loops of each escutcheon is a boss or short stump of metal

crossed vertically by a low ridge. The wall of the vessel must have been

perforated to permit mechanical locking of the escutcheon to the vessel. Thecopper inlay on the exterior goes underneath the edges of both escutcheons

which indicates that they were fixed in place after the copper was inlaid. Onhu 15.103 (fig. 84) the escutcheon also extends over the edge of the silver

inlay. All of these fixed-on escutcheons fit the contour of the vessel surface

nearly perfectly; there is no indication of joining with either soft solder or

hard braze metal.131 Perhaps these escutcheons with stems were placed at

proper positions in a mold, and the vessel was cast to them. If the stem was

cut off on the inside not too close to the vessel wall, it would leave, after

finishing, a boss like that on the inside of hu 61.32 mentioned above. 132

The escutcheon and ring handle on tun 38.7 (fig. 87) is still different.

The mask is cast onto the already cast and inlay-decorated vessel; the en-

gaged nose ring (proboscis), however, is made from a second casting of

metal which overflows the mask proper and enters a perforation in the side

wall to form a rivet-like locking device on the inside.

14. Swing handles. The only type of vessel on which an articulated handle

or bail is used is the covered wine jar or yu. There are two varieties of

swing handle; the ends of the simpler kind are provided on the inside with

sockets which fit on trunnions cast on opposite sides of the neck of the

vessel as was done with yu 40. 1 1 { fig. 88). The bail was probably just sprung

into position over the trunnions. The linked type of handle, however, is

more complex. The vessel neck is provided on the long axis with two cast-in

loops in the shape of a letter "C." The bails terminate in round eyes which

pass through the loops and provide the articulation. Often on the underside

of the eyes of the bails there is a short section with rough cast surface

bounded at either side with a ridge that cuts across the circle, not as a radius,

131It has not yet been possible to secure for the Freer Study Collection a vessel or fragment which could be

sectioned to show just how the escutcheon is joined.

132 In the Freer and in other collections, there are a number of detached escutcheon and ring handles of the

same type, which apparently served as handles of wooden furniture or of coffins. Almost invariably they are

made with a stem up to 5-10 cm. long which sticks straight out from the back.

88

Figure 84

Pien-hu 15.103. The handle escutcheon is cast separately

and secured to the vessel probably by a lock-on casting

technique. The escutcheon edges extend over the silver

inlay.

Figure 86

Fang-hu 61.32. One of a pair of escutcheon and ring

handles which are cast separately and then attached to the

vessel probably by casting on.

Figure 85

Fang-hu 56.15. The escutcheon appears at first glance to

be cast as an integral part of the vessel, but close examina-

tion shows that a seam exists around the edge, which

indicates they were cast separately and joined together.

Figure 87

Tun 38.7. The escutcheon is made in a peculiar way. Theanimal mask is separately cast and fixed to the vessel. Thenose loop is separately cast over the mask and is locked

mechanically through a hole in the side of the vessel. Theillustration at the right shows how the loop is locked to

the inside.

89

Figure 88

Yu 40.11. The swing handle is sprung over trunnions in-

tegrally cast to either side of the vessel. A band of decor

encircles the trunnion.

Figure 89

Yu 11.55. The ring of the bail of this archaistic bronze

was cast around the side loop. The narrowed portion at

the bottom of the ring with roughened surface probably

indicates the location of the ingate of the sprue.

Figure 91

Figure 90

Figures 90 and 91. Yu 30.26. Detail of the articulation of the handle. Figure 90. Common to many of the

swing handles of this type of vessel are the two vertical ridges that cut across the circular eye on either side

of the engaged loop (only one shown by arrow), not as radii but as parallel chords; they seem to mark the

edges of the sprue that served the handle mold. Figure 91. A radiograph of the articulation of the swing

handle shows that the closure of the eye is not perfectly round.

90

Figure 92

Yu 60.20. The upper surfaces of some of the swing

handles on yu are fully decorated. The sunken decor

on this handle is divided into four distinct sections

separated by cross marks that appear to be joins of

pattern units.

Figure 93

Huo 33.2. Mold marks on the underside of the link

that joins vessel and lid indicates that the entire

articulated assembly was put together by a casting

operation.

but as a cord parallel to the handle sides. Good examples are yu 11.55

{fig. 89) and 30.26 (figs. 90 and 91). At first it was thought that the eye was

made U-shaped to slip over the loop and that the eye was completed by a

supplementary casting operation, but careful examination of several of the

eyes reveals no sign of after-casting; the handle eyes are continuous metal.

This evidence indicates that the bail was independently cast on to an already

fabricated vessel and that the unfinished arc of the eye shows the location

of the ingate of the sprue of the special mold that was used to form the bail.

Chemical analyses of the swing handle of yu 40.1 1 and 30.26 (see Table B,

Appendix I) show that the composition of each is sufficiently different from

the vessels they belong to and indicate that the two members were poured

91

from separate alloy melts, hence they were formed in separate casting

operations.133

There are some other puzzling aspects of swing handles. For example,

handles of some yu are richly decorated in sunken decor, and the long decor

pattern is often divided into three or more distinct sections. On the handle of

yu 60.20 {fig. 92), it is difficult to tell whether these are mold sections or

pattern sections.

15. Link hinges. The covers of the spouted vessels called huo are some-

times hinged to the body with one or more chain links. The links on huo

33.2 (fig. 93), have vestiges of mold marks so it is probable that after the

vessel and cover were cast a separate mold was made to form the link to

join vessel and cover in one pouring operation.

16. Flanges and other appendages. The notched flanges on many Shang

and early Chou bronzes are cast as integral parts of the vessel. Very fre-

quently vestiges of mold marks show on the edges or in the middle of the

flanges. There are two instances, however, where the flanges were cast

separately and fixed to the vessel. On chia 07.37, two of the paired flanges

are cast as one with the vessel, but curiously the other three apparently are

precast and the vessel is cast to them. Furthermore, on the inside, directly

opposite each floating flange is a thickening of metal to provide proper

anchorage for the flange. The alloy of the floating flanges is yellower than

the body metal and hence probably of different composition. The handle of

this vessel is also cast separately; but unlike the flanges, it is cast on to the

vessel. Some of the heavy flanges of kuei 31.10 (fig. 94) are made in the

same way. The flanges of the small hu 49.5 are also made separately, and the

vessel is cast to them. One of the precast flanges of the cover actually breaks

through the inner surface {figs. 95 and 96). There is no explanation for this

peculiar way of making flanges. 134

There are many other kinds of appendages, whose purpose seems chiefly

to be decorative. There are spikes, knobs and animal masks in the half round

and in the full round, and there are capped columns peculiar to chueh and

chia. Some are solid and others are clay cored; some are integrally cast with

133 Noel Barnard in his extensive review of Cheng Te-kun's two books, Prehistoric China and Shang China, andof William Watson's China, attempts with aid of a drawing to show the construction of a special piece mold for

casting bail handles on vessels. He believes that pre-cast handles with partially open eyes were placed in posi-

tion over the side loops of the vessel and then by means of an auxiliary mold placed over the loops the eyes

were closed and the handles secured by a second pouring of bronze. This theory is not supported by observa-tions on swing handles of Freer vessels described above.134Cyril S. Smith, "Materials and the development of civilization and science," suggests that flanges did not

originate as a decor feature but evolved from the fins that are inevitably found at the juncture of piece moldscaused by metal running into the narrow space at the mold joins. He says that the earlier Chinese artisans ex-

ploited this defect to make it into a basic feature of the design, Flanges also probably facilitated the alignmentand register of the mold pieces. It should be pointed out, however, that pre-An-yang vessels often do not haveflanges or only small ones.

92

Figure 94

Kuei 31.10. The four pairs of ornate flanges midwaybetween the handles are precast and the vessel is cast to

them. This is indicated by the overlap of vessel metal onto

the flange and the seam where the overlap terminates.

The eight flanges on the foot, however, are cast integrally

with the vessel.

Figures 95 and 96. Hu 49.5. Inside of cover (about

natural size). The arrow shows the outline of one of the

precast flanges protruding through. The narrow seams

between flanges and cover show plainly on the radio-

graph.

Figure 96

the vessel and others are cast-on. Most extraordinary are the four protrud-

ing arms, already referred to (fig. 82), that stick out from the sides of yu

30.26. These are cast separately with clay cores and are fitted over bosses

that were cast integrally with the vessel side and secured at the point of

contact with hard solder. Likewise radiographs show that the separately cast

animal heads on tsun 51.19 (figs. 108 and 109), are fixed to bosses cast into

the shoulder of the vessel.

1 7. Capped posts of chiieh and chia. The small capped posts that stand

erect on the rims of chiieh are usually cast solid and integrally with the vessel,

but the larger caps and posts of the vessels called chia are quite another

matter. X-radiography reveals that the caps and posts of the three chia in

our collection are quite dissimilar in construction. The bird-shaped caps of

93

Figure 97 Figure 98

Figure 99 Figure 100

Figures 97-100. Chia 35.12. Figures 97 and 98 are

details of one of the capped posts seen from the inside and

outside. Faint join lines (shown by arrows 1 and 2) in-

dicate the caps were cast separately and joined to the posts

by liquid metal run between them. A mold mark in the

joining metal (arrow 3) indicates an auxiliary mold wasset on the post for this purpose. Figure 99 is an all-over

radiograph which shows further evidence of joining of

caps to posts; it also shows the holes on the inside of the

legs which probably result from spacers or supports

which connected the outer mold with the leg cores during

casting. The thin jagged white line that crosses from rim to

handle is a modern repair. Figure 100, a radiograph of the

same cap and post shown above, shows that both membersare cast separately and cored

;they have been joined with a

secondary run of metal which shows principally at the top

of the post.

94

Figure 101 Figure 102

Figure 103 Figure 104 Figure 105

Figures 101-105. Chia 23. 1 . Figures 101 and 102 are details

of one of the capped posts seen from inside and outside.

Absence of seams indicates that these features are cast

integrally with the vessel. Figures 103, 104, and 105 are

radiographs which show that the earthen cores in both

posts and both caps are continuous; they also show that

cores are crossed by metal pins which possibly are a

variety of chaplet. Figure 102. Ends of the pins are visible

in the exterior of the sidewalls (arrows). Figure 105. Onecap especially has a sponge-like structure caused by gas

bubbles trapped in the molten metal.

95

chia 07.37 are cored but the thin posts are solid and are cast as one with the

vessel; they terminate in a horizontal platform shaped to the underside of the

bird (see Chapter VII, fig. 212). A cone-like structure, whose purpose is

unknown, projects upward from the platform into the earthen core. Close

examination shows a narrow seam running around the edge of each bird

separating it from the platform. There is no evidence of hard solder; just

how the union was effected is not precisely known but it is probable that

the join was made by some variant of interlock casting. 135

X-rays reveal that an even more complicated structure exists in the caps

and posts of chia 35.12 (figs. 97-100). Both the cap and stem of each post

are earthen cored and the interior side walls of metal are straight and

smooth. Significantly, there is a faint horizontal join line running around

each sq uare post about a third ofthe way down from the underside of the cap.

A join line also runs around the lower horizontal edge of each cap. Themetal between the two join lines is apparently a special pour run in to join

the precast cap and vessel. There is faint evidence of a mold mark in the

joining segment running between the top of the post and the edge of the cap

which indicates that perhaps an auxiliary mold was built on top of the posts

to contain and shape the joining metal. The structure shown in the X-ray

suggests that the core of the cap was hollowed out to leave a square plug

of core sticking down. We can further conjecture that the core in the open

top of each post was hollowed out so that the cap core stem could be fitted

down into it thus serving to keep the posts and caps aligned during the

joining operation. On cooling, the joining metal would lock with the under-

side of the cap and the hollowed-out post. The horizontal seam seen in

the radiograph was probably caused by imperfect wetting of the inside walls

of the post by the joining metal, in turn caused by oxide dross and trapped

air at the interfaces. The caps on a chia in the Minneapolis Museum of Art

(Karlgren, Pillsbury Catalogue, pi. 14) seem to be joined in the same way.

The third chia 23.1 is made quite differently. The caps and posts, which

are earthen cored just like the legs, are cast integrally with the vessel (figs.

101 and 102). No sign of a seam shows in the X-ray or on the underside of

the cap; the earthen cores of cap and post seem continuous. One cap is

full of gas holes but the metal of the other is quite sound. Perhaps the

sounder cap and post served as a down-sprue and the porous cap as an up-

sprue. These caps have another unusual feature : X-rays show within the core

of each two opaque horizontal lines about 2 mm. wide running across the

135 Barnard in a book review of Cheng Te-k'un's Chou China in Monumenta Serica, p. 326, figure five, no. 1,

describes a broken cap of similar construction on a chia in the Sackler collection. He conjectures that the capwas first cast with its core, then the core was excavated. Just after the vessel was poured the cap was placed

over the molten metal in the sprue opening at the top of the post so that the molten metal could flow into the

excavated space and when cooled lock the cap to the post.

96

Figure 106 Figure 107

Figures 106 and 107. Yu 30.26. Figure 106. The knob is cast around a clay core. Two of the three holes in the

underside that lead into the core are visible (arrows). Figure 107. The radiograph shows that the core is

conically shaped; also it extends down into the stem.

core; the ends of these lines coincide in position with four small triangular

plugs in the cap walls which look like ordinary chaplets (figs. 103-105).

These inserts are interpreted as pins, perhaps a special kind of chaplet, that

join the opposite sidewalls of the cap. The two pins are at right angles to

each other but at a 45 degree angle to the axes of the vessel. Also two cross

pins, one above the other, run through the core of each post parallel to the

vessel handle. This is the first time we have seen pins or pin-type chaplets in

any of the ancient vessels.

The difference in casting techniques employed to make the same feature

in these three vessels drives home the point that each foundry in ancient

China had its own workshop methods which varied tremendously with time

and geographical distance. It also warns us to be wary of generalizations

based on isolated instances.

18. Knobs. The devices on the covers of yu which serve as hand grips are

generally of two types. One is simply a sort of flaring bowl set in the top

of the lid. Equally common is a round knob, usually fluted, that is joined to

97

the cover with a short stem. It has been discovered that the majority of these

knobs are cored with clay; moreover, on the underside of the knob between

the stem and the outer edge are three symmetrically placed small holes that

lead into the core as are plainly shown in figures 106 and 107. The reason for

the clay core is not known, but it is possible that it served to prevent excess

shrinkage as explained in section 22 below. It is hard to believe that such

small cores served to save precious bronze metal because the saving wouldbe negligible. There are about a half-dozen of these clay-cored knobs on yu

in the collection.

19. Contact fusion. It has been suggested that many of the legs and handles

and appendages which show no sign of seam or join with the vessel were

actually precast, but union of the pre-cast member with the main memberformed in the second pour is so perfect that the line of join is imperceptible.

This implies that the heat from the second pouring permits perfect fusion of

the melt of the second member with the member of the first pouring. Noevidence of such perfect fusion was found in any of the vessels examined,

and no evidence even of incipient fusion has come to light. Unless the tem-

perature of the mold was raised to near the melting point of the metal of

the precast member, no fusion could be expected when fluid melt met solid

metal since it takes time for heat to conduct and to bring solid metal to the

point of fusion, especially if the solid metal member has size and weight.

The heat of the melt was not sufficient in most cases to cause melting of

more than the edges of the small chaplets in the walls of vessels, and these

have infinitely less heat capacity than vessel members. 136

There is good evidence, however, that most handles and legs were cast

with the vessel. The broken-off handle of a p'an in the study collection

(SC537) is a good example. When first examined, it was thought that the

tie bars between handle and vessel edges might be precast and placed in the

mold in such a way that the main pouring of metal would engage them.

When the handle was cut in two, however, it was quickly seen that handle

and vessel are cast as one {figs. 110 and 111).

20. Chaplets. One of the most interesting features of the early vessels is

the existence on the vessel sides and bottoms at more or less evenly spaced

intervals of irregular quadrilateral insets of metal which seem to have been

136 Herbert Maryon speaks of "burning together," "fusion welding" and autogenous welding as methods em-ployed in both ancient and modern times to join members or to make repairs; but he makes clear ("Metalworking in the ancient world") that the joining parts must be raised to high local temperature. This could bedone only if the metal worker could bring to the critical spot a considerable bulk of molten metal and keep it

hot long enough to fuse the surface of the spots to be joined. In modern times this can be done with electric

arcs, acetylene torches and thermite, but in ancient times welds in bronze could be made only by surroundingparts to be joined with a special mold with overflow hole which permitted metal to be poured across the join

for a long enough period to melt the join ends. No evidence of such joining techniques were found in the Freer

bronzes examined.

98

Figure 108

Figures 108 and 109. Tsun 51.19. Figure 108 shows that

each of the animal heads is cast separately and joined

to the vessel mechanically. Figure 109. The radiograph

indicates the head is secured to an integrally cast boss onthe shoulder of the tsun, but the exact method of attach-

ment is not known.

Figure 1 10 Figure 1 1

1

Figures 110 and 111. P'an SC537 (handle). Figure 110.

On many flat vessels of this type, there is a tie bar

between the handle and the edge of the vessel. Figure 111.

A section through the handle shows that the tie bar is

cast as an integral part of the handle and vessel rim.

Figure 109

Figure 112

Tsun 54.122. In each of the undecorated areas that

separate the rising blade decor is the faint outline of a

squarish metal plug (arrows). The fact that these plugs

are located symmetrically about the vessel indicates that

they are chaplets placed in the mold to maintain a uni-

form thickness of the mold cavity.

Figure 113

Figures 113 and 114. A chaplet in the side of a p'an in

the British Museum (1953/5-11/1) bears a portion of lei-

wen decor. Figure 113. It apparently was made from ascrap piece of broken or discarded vessel. Macrograph x 3.

(Courtesy of Trustees of the British Museum.) Figure 114.

Figure 114

The outlines of the /e/-vt'e«-bearing chaplet are barely

visible on the vessel interior lying midway between twocast-in characters of the inscription. Macrograph x 3.

(Courtesy, Trustees of the British Museum.)

99

used as spacers or chaplets to keep the outer mold and the core separated at

a proper and even distance during the pouring process as seen in tsun 54. 122

{fig. 112). Because of their symmetrical disposition, chaplets are readily

distinguished from repair patches. They are usually made of a bronze alloy

which, in some cases, does not differ markedly from the alloy of the casting;

in others the copper content runs higher. Analysis of a chaplet from p'ou

13.30, shows: Cu 83.0%; Sn 15.7; Pb none; total 98.7, which is very near

the composition of the matrix metal. 137

There is no evidence that chaplets in the form of pins or nails as used

today in lost-wax casting were known to the ancient Chinese. 138 (A possible

exception is the thin metal pins revealed by X-rays which cut across the

clay-cored capped posts of chia 23.1 shown in figures 100-103.) It is a

mystery how flat chaplets were employed. They may have been fixed to the

core by a temporary adhesive or by some such simple means.

It is difficult in most instances to tell how the chaplets were made. Onegains the impression they were usually produced by simply breaking uppieces of discarded cast bronze. The chaplets in the p'an (SC537) mentioned

in section 19 are made of such bronze. Thin cast bronze could have been

broken into small squarish pieces using a tool something like a cold chisel.

That the chaplets were made of cast bronze scrap is supported by the ob-

servation that they occasionally bear on their surface bits of the familiar and

well-known lei-wen pattern. A lei-wen chaplet shows clearly in the side of a

p'an(No. 1953/5-1 1/1) in the British Museum (figs. 113 and 114). Another

shows in the lower body ofa kuei (No. 1956.854) in the Ashmolean Museum,Oxford (fig. 115). Still another occurs under the body bulge of our hu 48.1

(fig. J 16). There was probably no rigid rule about composition or fabrica-

tion of chaplets since they served only a temporary function and were ex-

pected to be invisible or nearly so. The great majority we have seen range

from 3 to 10 mm. across, and most are square or roughly square in shape.

It was observed in kuei 41.8 that some chaplets are partly but not fully

cast over by the vessel metal - a feature indicating that their thickness

was slightly less than the width of the casting space. This is clearly shown in

a cross-section of one of the chaplets observed in p'an SC537 and described

in detail in Chapter V.137Yetts (Eumorfopoulos . . . , p. 37) had observed chaplets but he thought they were plugs inserted into the

side walls after casting to conceal holes left by pins (perhaps of iron) used to secure the cores in proper rela-

tionship to the mold.138The foundry term "chaplet" is defined as "any of various metal devices for holding a core or section of a

foundry mold in place" (Gove, Webster"s third new international dictionary . . . , p. 376). H. Garland and C. O.Bannister in their little book Ancient Egyptian metallurgy, say (p. 39) that there is evidence the Egyptians used

"cross supports" or "struts" made of iron wire for this purpose. Chaplets in the form of pins are shown in a

diagram employed by R. F. Tylecote (Metallurgy in archaeology, p. 122) to show how molds were constructed,

and Jack C. Rich (The materials . . . , p. 149) has a splendid diagram showing the use of "metal supporting

pins" in a mold made for lost-wax casting.

100

Figure 115

A chaplet with lei-wen imprint also shows under the body

bulge in a kuei (1956.854) in the Ashmolean Museum.

(Courtesy, Department of Eastern Art, Ashmolean

Museum, Oxford.)

Figure 119

Figures 117-121. Huo (privately owned). Figure 117. This

small Han dynasty bronze tripod vessel is of exceptional

interest because of the large number of square metal insets

(chaplets) that are found on its surface. Figure 118. Therandom distribution of the chaplets is best seen on the

underside. Figure 119. The chaplets on the underside of

the huo still show parallel grooves which indicate they have

Figure 116

Hu 48.1. A chaplet with lei-wen imprint shows under the

body bulge of this large vessel. There is a corresponding

chaplet, but without lei-wen, on the opposite side.

Figure 120 Figure 121

been cut off with some kind of saw. The same parallel lines

show on the inside (opposite side) of the same insets.

Figures 120 and 121 . The chaplets on the upper part of

the vessel above the encircling rib are barely visible,

apparently the result of better finishing in this area. Details

about x 3. (Courtesy, Martin Nadel, M.D.)

101

Chaplets are visible in somewhat more than a third of all the Freer

vessels. They occur in some of the Shang bronzes, and seem to increase in

number through the Chou dynasty. Squarish chaplets are found in great

numbers in the sides of three of the Han dynasty lien (23.2, 51.2 and 51.5),

but apparently are absent in a fourth one (46.11). Strangely, chaplets seem

to be entirely missing in those vessels classed as "archaistic," and also in the

ones recognized as ming-chH.

A small three-legged bronze vessel in a private collection in Washington

{figs. 117 and 118), has supplied much interesting information on chaplets.

The sides of this Han dynasty huo contain perhaps as many as three dozen

square chaplets. Those below the encircling rib and on the bottom are quite

plainly visible. Each chaplet shows on its inside as well as the outside sur-

face parallel grooves which appear to be saw or coarse file marks {fig. 1 19).

The level of some of the chaplets and the saw marks on them is below the

level of the vessel side, which indicates the chaplets were cut off prior to

placement in the molds and not after the casting was completed. The saw

marks on the inside surface are added evidence of this. Whereas the chap-

lets on the underside are plainly visible, those on the upper part above the

encircling rib are barely visible {figs. 120 and 121). Apparently the upper

part of the vessel was given a final finish while the lower part was left in its

as-cast condition. Since visible chaplets in the upper part are fewer than in the

lower part, it is possible that many here are completely concealed.

It is difficult to give exact figures on the occurrence of chaplets because,

even if present, they may be overlooked for several reasons : (a) they may be

covered over by the melt; (b) they may have been concealed in the finishing

process; or (c) they may be entirely covered over by corrosion products. Onsome vessels they are not revealed by a seam but only by a color difference

in the patina. There is slight evidence that during the pouring of the molten

bronze into the mold, the chaplets may partially melt at the edges and hence

merge imperceptibly with the bulk pour if the temperature in the mold is

high enough and sustained (see Chapter VI I,figs. 218-220).

Usually, but not always, the chaplets are symmetrically disposed. On the

side of one tsun (54,122) mentioned above {fig. 112), they are located on the

quarters in the narrow spaces which separate the rising blades of the upper

decor. On the rectangular base of kuei 38.20 {fig. 122), they are located

at four corners in the plain band that frames the base. On fang-ting 50.7

{fig. 123), a single chaplet is in the middle of the undecorated panel on each

of the four sides. On some ting and kuei, they are symmetrically placed

under the body bulge. In the bottom of chia 35.12, a chaplet made of reddish

alloy was located in one corner directly over an integrally cast leg with clay

102

Figure 122

Kuei 38.20. The detail of one side of the square base

shows four symmetrically placed chaplets. Two of them

can be seen in the plain band that frames the tao-t'ieh,

just above the large hooked horns. The other two are in a

corresponding position on the band that edges the

bottom. Each of the four sides is the same.

Figure 123

Fang-ting 50.7. The detail shows a chaplet in the center of

one side of the square vessel. Each of the four panels has

the same squarish chaplet.

Figure 125

Kuei 11.38. Detail of the interior bottom showing a

chaplet in the area of the inscription on which a portion

of one of the characters is incised or engraved. See also

Chapter VI,figures 191 to 194.

103

Lien 51.2. The detail shows square chaplets in the side-

wall (arrows) revealed by the slightly lighter tone of the

patina that conceals their edges.

Figure 127

Lien 51.5. The detail shows three squarish chaplets in the

sidewall. They are partially concealed by decor lines en-

graved subsequent to casting. Note that the chaplets are

located between the relief figures where the walls of the

vessel are thinnest.

core. There is evidence of similar chaplets in the other corners. Squarish

chaplets, rather evenly spaced, are easily visible in the plain band that sur-

rounds the lid of the thinly castfu 13.14 (fig. 124).

Often in vessels which have inscriptions on the inside bottom, the chap-

lets are found just outside the inscription area. In kuei 11.38 (fig. 125),

chaplets are actually crossed by characters of the inscription- a circumstance

which will be discussed in Chapter VI. In the three Han dynasty lien men-

tioned above (figs. 126 and 127), the chaplets are random. The chaplets in

the huo mentioned above are also randomly disposed.

It was first held by the writer, and it has been suggested by others, that the

metal inserts are not in themselves chaplets but are plugs put in during

finishing operations to fill holes made by non-metal chaplets or other de-

vices to separate mold and core prior to casting, but all the evidence seems

to deny this. In all of the Chinese bronzes examined there is no evidence at

all that the symmetrically placed inserts are plugs. There are no signs of

hammering or peening of the inserts. There are hardly any drop-outs as one

104

Figure 128

Ku 17.202. When it came into the collection both top and

foot of this vessel were already broken off leaving the

high-set bottom exposed. Cleaning of the metal revealed

an irregular hole now partially filled with cuprite.

Originally this may have been the position of a chaplet.

Figures 129-131. Cross sections of three ku in the Freer study collection (fig. 129, SC528; fig. 130, SC529;and fig. 131, SC530) show the presence of chaplets in or near the center of each of the narrow high-set

bottoms. The macrographs (about x 7, x 6 and x 6 respectively) below each photograph show that the matrix

metal is cast around each chaplet.

105

might expect of plugs inserted in thin-walled vessels. In some cases, the

metal of the vessel clearly has run into depressions in the sides of chaplets

(see Chapter V, fig. 165). Furthermore, all the evidence thus far indicates

that some of the chaplets were made of cast high-tin bronze which wouldnot be suitable for hammering in.

21. Chaplets in ku. In the several ku in the Freer collection, no chaplets

have been observed in the vessel sides; but there is a single chaplet in the high-

set bottom, a feature shared apparently by most vessels of this type. Examin-

ation of the underside of the bottoms of ku 40.3 and 53.9 reveals in each a

squarish chaplet which is redder in color than the surrounding metal. In others

partial cleaning away of the corrosion products underneath reveals a brownpatch which is taken to be evidence of a chaplet. More drastic cleaning of the

underside of the bottom of ku fragment 17.202 (fig. 128) shows a hole in the

center now filled with reddish-brown cuprite which appears to have re-

placed a chaplet long ago fallen out. It seems that chaplets were necessary

to separate the upper and lower cores to permit the inflow of metal to form

the vessel bottom. By good fortune there are available, in the study collec-

tion of the Freer, three fragmentary and discarded ku which could be used

for experimental purposes. Two of these are heavily corroded early types,

dating possibly from the Shang or Chou dynasties; but the third is obviously

much later and may be an archaistic piece made in the Sung to Ming periods.

Each of the pieces was cut lengthwise with a band saw. This revealed immed-

iately that the high-set bottoms were cast as an integral part of the vessel (figs.

129-131). It revealed also that each bottom bears a chaplet of reddish metal

of varying diameter. In the oldest and most corroded piece, the chaplet is

about 4 mm. in diameter, but in the other early piece it is 12 mm. ; and in the

late bronze the diameter is about 15 mm. The reddish color of the chaplets

in contrast to the yellow color of the vessel alloy indicates higher copper

content and hence higher melting point than the vessel alloy. Chemical

analysis of samples of two of the vessels and of their chaplets shown in

Table VIII confirm the high copper and low tin content of the chaplets.

When these high copper content chaplets were first encountered it was

thought that the metal founders might have used them intentionally because

they have a higher melting point than the vessel metal, hence there was no

danger that they would melt out and cease to be useful during the casting

process. Later, however, it was found that sometimes, as previously men-

tioned, chaplet metal differs little in composition from matrix metal, hence

it appears that the Chinese founders had no fixed rule about their compo-

sition. We can conclude from this and from the discovery of chaplets with

lei-wen designs, mentioned in section 20 above, that they simply used metal

106

TABLE VIII

Analysis of chaplets in bottoms of ku

Vessel Cu0/10

Sn

%Pb%

Total

%Remarks

SC529Body metalChaplet metal

68.6

85.1

10.2

2.4

18.6

2.9

97.4

90.4

Some Pb lost

Single analysis ; metal heavily

oxidized

SC530Body metal 69.4 11.0 15.9 97.1* Some Pb lost

Chaplet metal 96.7 1.4 0.4 98.5

* Total includes Fe 0.4%; Ni 0.4%.

at hand which could easily be broken up into convenient size. It is curious

that Wan Chia-pao in his "The Process of Casting the Ku-beaker" madeno mention of these chaplets.

There is no certainty that all ku bear chaplets. A privately owned example,

which may be a Sung copy of an archaic type, has none. This same late

copy has none of the perforate crosses usually seen in the early ku. When an

ancient but intact ku is examined it is usually difficult to tell if there is a

chaplet in the bottom because corrosion products invariably obscure the

surface; and also, because of the distance of the bottom from either end,

it is difficult to probe the area. The chaplet, moreover, may not be exactly

centered.

22. Earthen cores. Frequent mention has already been made of the clay

or earthen cores in legs, handles, knobs and other appendages of Vessels and

as already explained in section 5 it seems that their main purpose was not to

economize on costly metal, but to reduce shrinkage of the metal when it

cooled from the melt. Cores may have been used for additional reasons

which we cannot appreciate.

The earth of the cores is fairly fine grained and uniform in texture; it is a

sort of low-fired earthenware pottery or terra cotta usually soft enough to be

heavily scored or dug out with a pocket knife. The harder varieties are

reddish and the softer ones grayish, perhaps indicating different degrees of

firing. How much of the hardness is caused by pre-firing of core, if any, and

how much is due to contact with molten metal at time of pouring is difficult

to say. The presence of carbon accounts for the dark gray or black of somecores.

During the course of this study samples from about 25 different cores

107

Figure 132 Figure 133

Figures 132 and 133. Ting SC587. Figure 132. This view shows the mixed minerals which make up the earthen

core of the leg of a vessel. In figure 133, taken with crossed polaroids, the sharply angular character of the

quartz particles is clearly shown. These photomicrographs were made from a thin section of the compact core

mounted in bakelite. 210.

Figure 134 Figure 135

Figures 134 and 135. The size and character of particles from a crushed fragment of a section from a baked

earthen mold (ROM No. 41) for casting bronzes are shown in these photomicrographs taken by transmitted

light at x 210. The high quartz content is obvious especially infigure 135 taken between crossed polaroids.

108

Quartz 3.34A

26 45 41 37 35 33 31 29 27 25 23 21 17 13 9 5degrees

d- spacing , 4.47A 12.35

A

. .,2.53

A

( Angstroms J

Figure 136

Chart 3. X-ray diffractometer analysis of three specimens of earthen core from bronze vessels (upper 11.41,

middle SC587, and lower 30.26) shows that quartz is the major component, and plagioclase felspar the next

most abundant mineral. Little clay mineral is present. (Tracings courtesy Department of Mineral Sciences,

Smithsonian Institution.)

109

were mounted for microscopic study at high magnification. In nearly all

the predominance of angular grains of quartz of quite uniform particle size

was noted (figs. 132-135). Other minerals accompanying the quartz are

plagioclase felspar, green hornblende, carbonate (probably calcite) andscattered grains of hematite, olivine, corundum, mica and others. Theinteresting fact is that there is little microscopic evidence of clay minerals139

or of clay minerals altered by firing in spite of the fact that the earthen

material does harden on firing. This is borne out by X-ray diffraction

analysis of samples of core material from three vessels, yu 30.26, ting SC587and ting 11.41 (fig. 136). It became apparent from further study of the

samples that the microscopic description of the core materials matches very

closely microscopic descriptions of the kind of earth called "loess" that

covers much of northern China where most of the bronzes originated.

Cressey gives us a map which shows that the chief concentration of loess

deposits are in Kansu province south of the Ordos desert, but they spread

southward to Shensi and Honan provinces and also eastward over Shansi

and Hopei where the loess is mixed with alluvium.140

In various reports that deal with the microscopic characteristics of loess

from China and elsewhere, the presence of a high proportion of angular

quartz particles is emphasized. George B. Barbour in an article, "The Loess

of China" says, "The true Chinese loess is a yellow-gray poorly-consolidated

loam deposit of the fineness of silt . .." (p. 456). After tabulating chemical

analyses of six samples of loess (p. 457) and illustrating particle characteris-

tics with a camera lucida drawing (p. 458), Barbour gives this mineralogical

description of loess: Excluding particles below 0.005 mm., "the average

diameter of 758 grains in this sample was 0.0124 mm. [Honan loess]. Thus

according to the U.S. Bureau of Soils classification, the bulk of the material

falls within the limits of silt (0.005-0.05 mm.). The outstanding features are

the angularity of the grains (which in many cases are practically free from

traces of rounding) and the surprisingly fresh condition of the mineral

grains, many of which are still almost unattacked by weathering ... it was

possible definitely to identify quartz, biotite, orthoclase and plagioclase

felspar, hornblende, carbonate, kaolinite and apatite in approximately that

order of abundance, together with some grains of one or two other minerals.

... To the naked eye this sample has a somewhat unusually high biotite

139According to Ralph E. Grim {Clay mineralogy, p. 1), "In general the term clay implies a natural, earthy

fine-grained material which develops plasticity when mixed with a limited amount of water .... Chemicalanalyses of clays show them to be essentially silica, alumina, and water, frequently with appreciable quantities

of iron, alkalies, and alkaline earths." Among soil geologists clay is a particle size term used to indicate soil

particles 2 microns (0.002 mm.) or less in diameter. In these discussions we often from habit and perhaps quite

erroneously refer to these earthen cores as "clay cores."14C''China 's geographicfoundations, p. 186.

110

content" (p. 459). A particle size analysis of two samples of loess from

Shensi province made by Professor W. C. Lowdermilk follows the general

description.

Particle size measurements were made microscopically on earth from

the cores of two vessels. The average diameter of 50 particles measured

from the core of the leg of ting SC587 is 0.0128 mm. and that of 50 particles

from core of yu 30.26 is 0.0085 mm. These values compare well with the

value of 0.0124 mm. given for the average diameter of 758 grains of Honanloess by Barbour.

The following table shows how the particles of the core material may be

classified in a quite arbitrary way

:

TABLE IX

Core materials: classification

Soil class* Diameter range Earthen coreSC587 30.26

Sand 2.0-0.06 mm. 2% 2%Silt 0.06-0.002 mm. 84 70Clay < 0.002 mm. 14 28

*On British standards metric scale for soil grades; see I. W. Cornwall, Soils for the archaeologist, p. 123.

TABLE XSize measurement of particles in earthen cores of seven vessels

Object Type Size of quartz

particles in

earthen core.

Average

Size of all

particles seen

in earthen core.

Average

Ratio of size

of quartz to

all particles.

30.26

30.26

40.11

60.18

61.30|SC587SC613f

yuyuyuting

ting

ting

ting

.0093 mm.

.0134 mm.

.0107 mm.

.0098 mm.

.0148 mm.

.0293 mm.

.0183 mm.

.0086 mm.

.0077 mm.

.0044 mm.

.0031 mm.

.0060 mm.

.0128 mm.

.0179 mm.

1.08

1.74

2.44

3.16

2.47

2.28

1.02

Average .015 mm. .0086 mm. 1.76

Soil from Fen RiverValley, Shansi Province

.0087 mm. .0038 mm. 2.29

Soil clinging to surface

of bronze vessel 24.1

1

.0153 mm. .0073 mm. 2.09

*The size of the particles was determined by using a micrometer eyepiece on a petrographic microscope. Theresults are numerical averages. In all cases, 90% or better of the quartz particles counted fell into the silt-size

range. In each case, twenty-five particles were counted, except for those objects indicated by a daggerin which fifty particles were counted,

f Fifty particles counted in each case (see above)

.

Ill

Figure 138

Figures 137 and 138. Chia 07.37. Figure 137. The tip of one of the legs has been repaired by casting on. Therepair was necessitated, apparently, by failure of the leg to fill out in the original casting. The leg proper is

clay cored. The bronze alloy used to make the tip has much lower tin content than the body metal. Figure 138.

The radiograph of the lower portion shows that the cast-on tip of the repaired leg is solid metal.

As already noted the most conspicuous mineral component of all samples

of earthen cores examined microscopically is quartz. The quartz particles

seem fairly uniform in size, and they form the bulk of the coarser particles

as is shown in Table X. The presence of so much quartz makes loess a

poor ceramic material so far as strength is concerned, but its presence im-

parts one quality probably much appreciated by the bronze founders - low

shrinkage on firing. The prime requirement for core material would be low

shrinkage, not strength. Another important feature of earthen cores as well

112

Figure 140

Figures 139 and 140. Fang-ting 50.7. Figure 139. Three of the legs are cast integrally with the vessel but the

fourth leg (upper right) was a later cast-on replacement as indicated by the spill-over of metal at the join with

the body (arrow). Perhaps in the original casting this leg failed to fill out properly. Figure 140. A radiograph

shows that the clay core of the cast-on repair leg of the fang-ting is improperly aligned. The other three legs

are solid and cast integrally with the vessel. (See also^zg. 214, Ch. VII.)

as the earthen molds themselves was porosity which permitted the escape

of gases during pouring of the metal.

Although the clay cores of legs, handles, and knobs often appear to be

entirely enclosed by cast metal, it is evident that the founders sometimes

provided holes for the escape of expanding gases which would fill the voids

of the porous earth during casting. Also, the inside faces of vessel legs are

often provided about half way up with small holes leading directly into the

core which perhaps provided additional means of gas escape. Triangular

113

legs of the larger chia are often ribbed or cast partly open probably for the

same reason and, as previously mentioned, some clay-cored knob handles

have three small vent holes just below the knob rim. The cores of kuei

handles are openly exposed in the handle channels. Presumably the center

cores and the outer molds also provided plenty of intergranular space for

expanding gases caused by the inrush of the main mass of molten metal at

the moment of casting.

According to J. C. Frye loess soils are used in America for makingbricks and tiles and other rough ceramic equipment. 141 It is well known that

much of the brick and tile of China as well as core materials for foundry

purposes has always been made from the "yellow earth" underfoot. Also the

black filling of the sunken decor found on many vessels and described in

Chapter IX contains a large proportion of angular quartz grains mixed with

carbonaceous material and resembles loess. Loess seems to have had

various uses in bronze vessel manufacture.

23. Ancient repairs. There are a number of vessels in the collection which

apparently did not emerge from the casting mold entirely whole and intact.

Accidents in the foundry sometimes produced faulty castings; some had

missing members and others were marred by incompletely filled-in sides. If

this happened, it was too costly to construct a new mold and to melt and

repour the vessel. Just as it was a routine operation to cast on legs and

handles, it was easy to cast on a missing member or to fix a patch. The tip

of one of the earthen core legs of chia 07.37 {figs. 137 and 138) apparently

did not fill out in the pouring. It was provided with a cast-on replacement

that was probably made directly after the vessel was first produced. Asshown in the figure the new tip is a little larger than the tips of the other

legs. It is also made from a different alloy, and strangely enough one which

contains a higher percentage of copper than the vessel itself and hence must

have a higher melting point. Analysis of the replacement gives: Cu 93.6%

;

Sn 3.3; Pb zero; total 96.9. On another vessel, a fang-ting 50.7, one entire

leg apparently failed to form on the first pour. Undaunted, the founder cast

on a whole new leg to complete the vessel (figs. 139 and 140).

In some instances the cast-on patches used to fill holes in the casting showthe stump of the sprue left from the pouring of the patch. In the irregular

patch on the underside ofp'an 1 1.35 {fig. 43), the stump is a low ridge about

1 cm. long and 3 mm. high. On chiieh 25.3 the cast-on round patch on the

bottom is about 2 cm. in diameter and has a sprue located in the exact center

(fig. 141); it is a fairly neat and well-executed repair. By contrast the patch

on the bottom of hu 24.12 is crudely made (fig. 142). Here a crack about

141Frye, "Ceramic utilization of. .

."

114

Figure 141

Chiieh 25.3. The detail shows an old mend in

the bottom made by casting on a patch. Thepatch flows over the edge on the inside. Thesprue of the patch shows plainly.

Figure 142

Hu 24.12. Detail of the underside showing a crude cast-on repair.

The short ridge in the center of the blob of metal at the left end of the

repair is probably the stump of a sprue.

115

12 cm. long has been filled with a ribbon of cast metal which was allowed to

flow in and lock into place by overflowing the edges inside and out. Theupper end of the patch has a short ridge-like stump, the remains of the

sprue. Repairs of this kind may have been made many years after the vessel

was fabricated, and the special molds used to make them were probably

quite simple and attached directly to the vessel. Contemporary repairs of

this kind made on ancient Chinese bronzes are not unusual. They have

been observed on early bronze vessels in other collections. Barbara Stephen

describes a chileh in the Royal Ontario Museum attributed to the pre-An-

yang period which apparently was repaired before burial by casting on the

spout and two of the legs, and there is also a repair of an apparently

damaged spot at its waist. Miss Stephen also describes a Shang dynasty

vessel of the type li which was damaged in antiquity and also crudely re-

paired before burial. The bottom and the legs were completely recast on to the

body and holes on the side were crudely repaired with poured metal patches.

Miss Stephen says that sections of the inner clay mold used for recasting

the bottom are still fixed to the interior. Four other early Chinese bronze

vessels at the Royal Ontario Museum have similar repairs.142

During the Roman period in Europe, and even going back to the Bronze

Age, there are many examples of the repair of casting failures employing

casting-on techniques. A notable one is the "Mantua Apollo" of the FoggArt Museum that was recently described by Mrs. Leonore O. KeeneCongdon. In the casting of this small Roman bronze figure, the right side

of the top of the head did not fill in with metal, but was left as a large

gaping hole. To effect a repair several holes were drilled around the edge

of the gap and then new metal was poured into a special mold fixed to the

top of the head so that the metal would run under the edges and up through

the holes to lock the patch securely in place. Hans Drescher has observed a

number of objects of the Bronze Age in Europe on which repairs were madeby casting-on or casting-in in a way quite similar to those that have just been

described in early Far Eastern pieces. 143 Drescher believes that in situations

where whole members were missing the member was modeled in wax and

surrounded with a mold as in cire perdue casting; but where holes in the

body were to be plugged or other simple mends to be made, a model was

not needed so that the mold could be shaped free hand. Various devices

were used to lock the patches on, like drilling holes, cutting grooves and

notches, or undercutting the edges of the parts to be mended.

Smaller flaws like gas holes in the castings of Chinese ceremonial vessels

142Stephen, "Shang bronzes with ancient repairs."143"Ausbesserungen. . .

."

116

Figure 145

Ku 51.18. Detail showing the inside (left) and outside of

two plugs fixed in the side to repair casting flaws. Theplugs were perhaps originally hammered in and smoothedflush to the surface, but they are now revealed by cor-

rosion, x 3.

Figure 147

Lead chiieh SC546. In this detail ( x 2) the raised edges

of the main decor elements are plainly shown. They in-

dicate that the design was first outlined in the mold with

a sharp tool, then the surplus material between the lines

was removed.

Figure 146

Tsun 55.1. The mold join mark which originally ran longi-

tudinally along the line separating the t'ao-t'ieh masks wasmostly eliminated on the waist and on the foot in the

finishing operation.

117

were repaired by simple methods. On fang-i 54.13, there is a crude metal

plug in the side of the vessel (see Vol. I, pi. 36). A repair plug in the side of a

chileh 25.3 {fig. 143) takes the shape of a bit of relief decor. The plug in the

spout of chiieh 54.15 {fig. 144) looks like a chaplet on the inside, but out-

side it is obviously a repair. A repair plug on the side of a ku 51.18 {fig. 145)

is so well made that it is only revealed by the lighter color of the patina indi-

cating that the metal of the plug differs in composition from the matrix metal.

It is difficult in many cases to tell if repairs are contemporary or made at a

much later date. We believe, however, that many are contemporary with

fabrication.

24. Finishing. The extent to which ancient bronzes were given additional

finishing treatment after removal from the mold must have varied consider-

ably. All classes of quality in casting and finishing were probably produced.

Some vessels were obviously indifferently cast; but most of the inscribed

bronze vessels, because they were intended for ceremonial purposes, were

well turned out. A few good early bronzes, which fortunately have escaped

eruptive corrosion in whole or in part, show surfaces nearly as smooth andhighly reflecting as the polished and finished bronze mirrors of later date.

In the Freer collection and in others, as previously mentioned, a few vessels

exhibit isolated patches of surface which are still glossy, metallic, and brownjust like well-cared for Italian Rennaissance bronze statues. There is someevidence from fine parallel lines on the surfaces that abrasion and polishing

were the final steps in fabrication. Differences between the corrosion

products formed in depressed surfaces and those on raised and exposed

surfaces is additional evidence of mechanical work (see Chapter VIII).

On the more highly finished vessels nearly all mold marks have been elimi-

nated, and vestiges of pouring gates have nearly always been removed

{fig. 146). The lead chiieh (SC456) {fig. 147) mentioned above perhaps gives

some idea how bronzes looked when they first emerged from the mold. The

sharp, raised edges on the main decor elements of the fao-fieh on this

vessel show that the design was first outlined in the mold with a sharp tool;

the surplus material between the outlines was then removed, but not to the

full depth of the outlines. They appear as ridges on both sides of most decor

elements on the piece. The same raised edges can be seen in reverse on a

mold fragment from the Royal Ontario Museum (Chapter II, fig. 10). Ona finished bronze these raised edges would have been removed by abrasive

tooling, as would the casting flash at mold joins which this vessel displays so

prominently.

25. Summary. The above observations extend our knowledge of vessel

making and lead us to an entirely new appreciation and an increased ad-

118

miration for the skill of the ancient Chinese foundryman. Unfortunately,

the evidence does not permit us to give definitive answers to all questions

concerning fabrication of the ceremonial vessels. Some things are sure :

(a) All of the ceremonial vessels in the Freer collection are formed by

casting, and there is abundant evidence both from mold marks on the

vessels and from surviving earthen mold fragments, that the casting

was done in piece molds. Apparently there was no intention of repli-

cate casting, and each set of molds was made to produce a single piece.

(b) The vessels of the Shang dynasty are generally cast in one piece, andoccasionally full round decor elements like animal heads were fixed

on mechanically. In Chou times, however, legs and handles of the

larger vessels were cast separately, and the vessels were joined to these

members by an interlock casting technique so that the joins are

almost invisible from the exterior.

(c) In addition to joining by interlock casting, early vessel fabricators

also sometimes joined handles and appendages with a "hard solder"

or by a technique closely resembling "brazing."

(d) The eyes of swing handles of yu were cast around loops which them-

selves were cast integrally with the vessel. Another type of handle was

cast separately and the socketed ends were sprung over trunnions.

(e) A surprising number of legs, handles, and knobs are cast with earthen

cores, and many of these cores are still in place. The core material seems

to be the yellow earth of north China often called loess. Moldmaterial also seems to be loess or silty earth, not plastic pottery clay.

It apparently shrank little when fired.

(f) A characteristic feature of many vessels is the presence in the walls of

metal chaplets which tend to be regularly and symmetrically disposed.

These appear to be spacers which served to keep core and outer moldapart during casting. They are quite unlike the pin or tie-bar type of

chaplet long employed in the West in cire perdue casting. Thechaplets are made of bronze of widely varying alloy composition. In

vessels of the type ku a single chaplet or a hole left by a chaplet is

almost invariably found near the center of the high-set bottom.

(g) The observations and conclusions in respect to use of piece molds and

interlock joining of separately made handles and legs is strongly

supported by the extensive finds of piece molds discovered in recent

years at Hou-ma in Shansi province and other sites in China. In the

technical details of the bronzes themselves we find confirmation of

several of the observations reported by Chang Wan-chung (already

mentioned) in respect to the Hou-ma finds. Chang shows pictures of

119

separate molds for legs and handles of ting. The use of tenons to

insure proper alignment of the mold sections is diagrammed. Hesays: "The process for casting a ting can be imagined thus: The body

of the ting and the fixtures were cast separately. First the handles,

legs and the lid were cast. A hole was provided in the outer mold for

the knob so that lid and knob were cast together. After the piece

molds for the body were assembled the handles and legs were in-

serted into holes provided in the outer body mold. (A clay body moldwith holes for setting in the handles was actually discovered; it is

shown in diagram fig. 11.) On the outside they put a coating of mud;and afterwards they take the mold and put it upside down in a hole

previously dug in the ground and attach a pouring cup through which

the molten metal flows into the bottom of the ting. The ting handles

and the ting legs join with the metal poured into the body of the

vessel. After a little, when it is cool, they take off the outer mold and

take out the inner mold (i.e. core) and rectify the seams on the outer

surface. Through such systematic steps the casting of the ting was

accomplished." 144

(h) There is no clear line of evidence, but it seems that at the close of the

Chou dynasty radical changes were introduced in the fabrication of

vessels. It is possible that these changes reflect the introduction of

cire perdue or some indirect method of casting from the West, but

this has not been demonstrated. Apparently piece molds were dis-

carded, but the traditional method of separating cores from outer

moids with metal spacers was retained. Old techniques of casting

sunken decor were discarded for newer methods of decor employing

engraving, chasing, and inlaying with precious metals. A tremendous

amount of work lies ahead before the complete story of the fabrica-

tion of ancient Chinese ceremonial vessels can be told. The answers

will have to await further studies in the laboratory and further

archaeological field work. It would help greatly if in the meantime

experienced sculptors and foundrymen could be induced to take in-

terest in these problems to an extent that they would attempt to

simulate the ancient techniques in their own workshops.

144Chang Wan-chung, "Hou-ma. ..." p. 42, translation by T. Katsuki.

120

CHAPTER V

METALLOGRAPHIC STRUCTURE

In the laboratory examination of any antique metal object, it is desirable

to study the physical structure as well as the chemical composition. Themicroscopic details of the crystals that compose the metal reveal much about

its technical history, including the manner of its casting, its forging, and its

heat treatment as well as both natural and artificial patination of the surface.

Such studies are carried out with a special microscope adapted to examine

polished surfaces of opaque materials by reflected light. A small cross-sec-

tion of the object being examined is cut out, mounted in a block of plastic

so that it can be easily handled, and polished with a series of graded abra-

sives to give an undistorted flat surface which is finally etched with acid or

other agent to reveal the structure as affected by the differing rates of

chemical attack depending upon the local composition, structure, or

orientation of the microcrystals. Although the crystal structure is usually

too small to be seen with the naked eye, some structures occur in metals on a

larger scale, such as the niye of the Japanese sword, the visible wood grain

effect (mokume) seen on some Japanese sword guards (tsuba), the texture

of Damascus steel, and the crystalline moire metallique of 19th-century

Europe. Crystals are often revealed on cast brass door knobs by mild

weathering.

It is also possible using mechanical or electrolytic methods 145 to polish a

tiny local area directly on the surface of a large or valuable object without

perceptibly damaging it, but deep surface corrosion limits the utility of this

on ancient metal objects.

Although the full interpretation of structures requires an extensive

knowledge of metallurgy, the commoner structures that appear in bronzes

are easy to recognize and interpret. 146 The structure of a piece of cast metal

is affected by the way it cools from a melt and so preserves a record of the

details of its history, for the crystals must be nucleated somewhere and growuntil they exhaust the metal in which they are growing or impinge upon

145See A. France-Lanord, "Le Polissage electrolytique. . ..".

l46por a discussion of the techniques of metallography, see G. L. Kehl, The principles of metallographiclaboratory practice (Third ed., 1949); for the interpretation of structures see any good textbook of physicalmetallurgy; also, C. S. Smith, "Some elementary principles of polycrystalline microstructure"; J. Nuttingand R. G. Baker, The microstructure of metals; Max Hansen, Constitution of binary alloys (1958) Cu-Pb pp.609-612, Cu-Sn pp. 633-638, Pb-Sn pp. 1106 1109. Also see Samuel L. Hoyt, "Composition of copper-tinalloys" in Metals handbook (1939 edition) pp. 1364-1366.

121

each other. In an alloy containing a second element, branched crystalline

structures called dendrites commonly form, and solidification usually

produces microsegregation between the branches. The metallurgist uses a

constitution (phase) diagram to represent the possible changes compactly.

This is a kind of map on which are plotted in terms of the composition of

the alloy and the temperature the areas in which solid or liquid exist alone

or in mixture, and the regions where the different types of solid crystal

form. The diagram for the copper-tin system, which includes the bronzes, is

shown in figure 148. It is at once obvious that the addition of tin rapidly

decreases the melting point of copper, i.e. an alloy with 25 percent tin melts at

a temperature almost 300°C (540°F) below pure copper.147 There are, more-

over, many distinct solid crystal forms or phases formed between copper

and tin. These phases, which are usually designated by Greek letters, differ

greatly in their properties, and most of them are hard and extremely

brittle. The diagram shows only the phases which exist when an alloy has

been held long enough at a given temperature to obtain equilibrium. Acasting usually cools too fast to allow equilibrium to be reached, but the

sequence of phases nevertheless in general follows the diagram. When a

liquid bronze containing about 15 percent tin by weight is cooled, the first

crystals of the solid phase called a will begin to form at about 960°C, and

it will contain only about 5 percent tin. Further cooling will deposit moresolid, and both solid and liquid will become progressively richer in tin - just

as salt is concentrated in water by the crystallization of ice. At a temperature

below 800°C, the last liquid which occupies the interstices between the

branches of the crystals solidifies in the form of the /S phase. With further

cooling j8 changes to the closely related y phase, which in turn undergoes a

profound change at 520 C when it decomposes completely into an intimate

duplex mixture of a and 8 called a eutectoid. Areas of such eutectoid can be

clearly seen in the photomicrograph (fig. 149). The more tin a bronze con-

tains up to about 25 percent, the more eutectoid will be seen in its structure;

beyond this 8 increases. The high-tin speculum metal of which Chinese

mirrors are made contains a high percentage of 8 phase. It casts easily and

takes a good polish but is brittle.

In a cast bronze the size and the regularity of the dendritic crystals vary

greatly with the conditions of casting. The slow solidification of undis-

turbed liquid metal gives large, well-formed dendrites (figs. 150 and 151).

During casting, however, unless both mold and metal are very hot the

metal may commence to solidify while it is still moving within the mold, in

which case many small crystals will form, having a less well-developed

147 See also Table, Chapter I.

122

Constitution (phase equilibrium) diagram for copper-tin

alloys showing the regions of stable existence of liquid

and various solid phases as a function of temperature andcomposition. The "liquidus" line marking the onset of

crystallization on cooling is the uppermost line. Althoughthe diagram calls for changes below 500°C, under nor-

mal conditions of metallurgical treatment the 8 phase

does not decompose, neither does the a phase deposit 8

or e. The dotted lines show the constitution to be expec-

ted under normal metallurgical treatment. {Metals

Handbook, 1948 edition, modified.)

Weight Percentage Tin

Figure 148

Figure 149

Microstructure of a polished and etched section of a cast Chinese bronze dagger, x 250.

The large areas in three shades of gray are differently oriented crystals of the a phase.

Embedded in them are irregular "pools" of duplex structure which were the last parts to

solidify; they deposited in the form of the y phase but on further cooling decomposed to

the oc-f 8 eutectoid. During cooling the y phase competed with gas cavities (black) and

with drops of lead (dark gray) for the interdendritic spaces in the a crystals. The faint

shading in the a crystals results from progressive enrichment in tin as solidification

proceeds. (Courtesy, Betty Nielsen, University of Chicago. Plate 2924, CSS.)

123

Figure 150

Figures 150 and 151. P'ou 13.30. Figure

150. Metallograph (x 120) of the edge of a

polished section of a metal specimen after

etching in ammonia and hydrogen peroxide.

Below the thin corrosion front of tin oxide

the dendritic structure of the unaltered core

metal is clearly shown. The absence of

Figure 151

scattered droplets of lead will be noted.

Figure 151. The same, x350. The etched

areas (depressions) are copper-rich a; the

light areas are tin-rich a and the clouded

areas are eutectoid. The black areas are

cavities.

geometric shape, although the microsegregation is not much changed. If a

bronze casting is hammered or otherwise deformed and subsequently an-

nealed at red heat, the metal will recrystallize completely and form newgrains having shapes (fig. 152) quite different from those in the casting.

As previously mentioned in Chapter III, lead was added to Chinese

bronzes for its effect in improving castability and because it makes the cold

metal more easily cut or worked with engraving or scraping tools; it also

confers a pleasant surface texture and color to the metal. The metallurgical

behavior of lead as an alloy in bronze is totally different from that of tin.

The constitution diagram of copper-lead alloys is shown in figure 153. Themetals are insoluble in each other in the solid state, and there is a large

region wherein two immiscible liquids are formed. On the cooling and

solidifying of a lead-bearing bronze lead is trapped in the growing crystals of

124

Microstructure of a bronze fishhook (from

Corinth, sixth century B.C.) showing that it

had been annealed and slightly cold-

worked. The metal had been annealed

several times in the course of fabrication

because the dendrites in the casting have

disappeared. The tin in the eutectoid has

diffused and has been taken into solution;

recrystallization has resulted in v. grains

that are approximately equiaxed. The fine

striations are slip bands, indicating that the

metal had been slightly cold-worked after

the last annealing operation. Etched, bright

field, > 500. (Courtesy, Katharine Ruhl,

Massachusetts Institute of Technology.)

Figure 152

copper-tin alloy or at their boundaries, and ultimately the droplets them-

selves solidify. Lead in minor amounts does not cause a substantial lowering

of the melting point as does tin.

Under the microscope the lead in cast bronze appears in two forms. If

there is a lot of it and if the alloy has cooled fairly slowly, some of the lead is

separated from the liquid alloy and appears as individual round droplets

{fig. 154). More commonly, the lead does not begin to separate before

solidification of the copper-tin crystals has commenced, and it then forms in

less regular patches, associated with the eutectoid in the interdendritic

spaces which are last to solidify {fig. 149). Any porosity arising from the

evolution of gas during solidification or from the inability of molten metal

to flow in to compensate for shrinkage will also occur in the same areas,

and it is sometimes difficult to distinguish lead from cavities of similar

geometry since lead is easily dragged out by the polishing operations.

Tsurumatsu Dono, in his investigation of ancient metal culture in the

Orient, already mentioned in Chapter II, shows a remarkable series of

photomicrographs of structures of early Chinese bronze weapons.148 Morephotomicrographs accompany the more recent article by Giichi Tanabe on148"The chemical investigation of the ancient metallic culture in the Orient," plates II-IX, XIII-XVI,XVIII-XIX.

125

Af/o Pb10 20 JO 40 50 70 90

1600

1400

1200

1000

eoo

600

400

200

L,q uid

//

//

/

—\

\-\

1

I

1 2i (en

./'quids

Pulsion)

\

\

\

1

1

1

\

\

1

\

1

»"'-^^ 1

7-4

1

Cu + Liq

*9S2° 2 Ll

—T1

quid iayi•rs

Cu +Liq

326°

Cu-h Pb

20 40 60W(% Pb

Figure 153

Constitution diagram for copper-lead alloy

showing the regions of stable existence of

liquid and various solid phases as a func-

tion of temperature and composition. The"liquidus" line marking the onset of crystal-

lization on cooling is the uppermost solid

line. The immiscibility of lead and copper

even above 327°C, the melting point of lead,

is clearly indicated. {Metals Handbook,

1939, via Colin J. Smithells, Metals

Reference Book, Vol. II (1967), p. 474.

Courtesy Butterworth & Co.)

Figure 154

Ku SC530. The lead in this bronze is not dissolved in the

copper but is distributed in little solid globules through-

out the alloy. Some evidence of the dendritic structure of

the copper-tin phase can also be seen in this metallograph

(bright field, x 100).

Figure 155

Chia 35.12. The crystal (dendritic) structure of cast bronze is

clearly revealed in the thinly corroded surface. The high-tin

constituent has been converted to white tin oxide. Photomicro-

graph made directly without polishing or etching the metal

surface (dark field, x 20).

early Japanese bronzes 149 though neither of these authors attempts to inter-

pret the structures of individual specimens shown. H. C. H. Carpenter does,

however, point out the relationships between metal structure and corrosion

products in the few fine metallographs that accompany his article on analy-

sis of ancient bronzes.150

As shown in Chapter VIII on corrosion phenomena, the surface of a

corroded bronze sometimes directly reveals the grain structure without

metallographic preparation. This is especially true of high-tin bronzes with

a smooth, gray-green tin-oxide patina like that on chia 35. 1 2 {fig. 155), where

ghosts of the dendritic structure appear as a result of the replacement of the

high-tin eutectoid with white tin oxide.

Although, as previously explained, no systematic program of metallo-

graphic analysis comparable to that of chemical analysis was set up for all

the bronzes, some special aspects of structure were studied; treated as case

histories they demonstrate the benefits to be gained from inquiries into

physical structure.

Metallographic study ofafragmentary early Chinese bronze vessel

Some years ago a broken and much repaired Chou dynasty hu came into

the possession of the late Charles Fabens Kelley, then Curator of Oriental

Art at the Art Institute of Chicago. Mr. Kelley in turn presented it to the

author for technical study with permission to analyze, to cut out sections and

to use experimentally in any way desired. This was a unique piece because

at that time there was little study material of this kind available to Western

scholars. 151 With the help of Mr. Daniel Cushing, metallurgical engineer of

Boston, a number of small pieces cut out from this bronze were mountedand polished. The results of metallographic examination were published

some years ago by the author using many of the metallographs prepared

by Mr. Cushing. 152 Some of these are in part reproduced here (figs. 156-

162). In the cross-sections the vessel wall shows three principal zones:

(1) in the center, small irregular areas of uncorroded metal, (2) an inter-

mediate partially mineralized zone, and (3) an outer completely mineralized

zone. Both the outer and intermediate zones have a layered structure

within themselves (figs. 157 and 158).

Microscopic examination of the core metal after etching showed directly

that the alloy is high in tin and that it contains scattered globules of un-

alloyed lead. (Chemical analysis of a sample of uncorroded core metal from

149"A study on the chemical compositions . ..," plates III and IV.

150 Carpenter "Preliminary report . ..

," plates II and III.151The repairs and false patina used to conceal them will be discussed in Chapter X.152Gettens, "The corrosion products of an ancient Chinese bronze."

127

Figure 159 Figure 160

Figures 156-162. Hu SC548. Small sections about 2 cm. x completely penetrated by corrosion. The narrow bright

1 cm. were cut from a fragmentary Chinese bronze vessel band above, and the broad band below are elemental

of the Chou dynasty in the study collection known as the copper redeposited from copper salts formed in the

"Kelley bronze." After mounting, polishing, and etching initial stages of corrosion. Going outward there are layers

they served for metallographic study. Figure 156. The of cuprite, then of malachite. Figure 158. Another area of

study specimen described below came from a fragment like the same section at x 50 shows some uncorroded metal,

this which originally was part of the rim of the hu. Figure the intermediate partially mineralized zone, and the outer

157. A view across a mounted section taken at low magni- completely mineralized zone. The dark gray band just

fication ( x 9.5) shows that in some areas the vessel wall is outside the zone of dendrites is cuprous chloride. The

128

Figure 161

bright reflecting porous layer beyond is red cuprite.

Figure 159. Metallograph (bright field, x 200) of a small

area of the uncorroded core after etching with potassium

dichromate solution. The spaces between the dendrites of a

bronze (light areas) are filled with fine-grained a+Seutectoid. The rounded black areas are voids formerly

filled with droplets of segregated lead. Figure 160. Another

area, same specimen at x400. Figure 161. In the inter-

mediate zone the main trunks of crystals of a bronze have

Figure 163

Figures 163 and 164. Fan SC537. Figure 163. The bottomof this broken and fragmentary Chou dynasty vessel con-

tains several chaplets barely visible to the naked eye. Oneof these was cut out from the matrix, and the specimen used

for metallographic study. Figure 164. The radiograph of the

broken vessel is quite revealing. In addition to showing the

presence of several squarish chaplets, it also shows that the

resisted attack but corrosion has darkened the high-tin

phase surrounding them ( X 100). Figure 162. This metallo-

graph at X 385 in the intermediate zone shows an irregular

island of redeposited elemental copper which apparently

has replaced a droplet of lead. The smaller islands of

bright metal are uncorroded a bronze; the black matrix is

corroded eutectoid. All photomicrographs are with bright

field illumination. (Figures 157, 158, 161 and 162, courtesy,

Daniel Cushing, Boston, Massachusetts.)

Figure 164

alloy on one side is much more dense to X-rays than on the

opposite which indicates that, in the casting, two batches of

molten metal of different composition were used. The lead

content of the denser metal is 18.3 percent, whereas the lead

content of the less dense metal is only 9.9 percent. The area

where the two melts meet is banded. (See larger view of this

radiograph in Ch. VI, fig. 197.)

129

the base showed: Cu 72.3% ; Sn 21.8; Pb 4.1 ; Total 98.2.) The a dendrites

are well formed and are interspersed at grain boundaries of a and 8 eutec-

toid (figs. 159 and 160).

Of chief interest is the intermediate penetration zone where the eutectic

has been converted to a black product leaving dendrites of a bronze un-

attacked. This is selective corrosion which results in a natural etch. Because

the black material is so intimately mixed with uncorroded a it was im-

possible at that time to isolate it. From a polished section, however, it waspossible to isolate chips free from outer corrosion products and to subject

them to X-ray diffraction analysis. The powder pattern showed mainly an

expanded copper lattice and faint lines characteristic of cassiterite, Sn02

;

and also one unidentified line. No lines of the copper chloride minerals,

which exist in the outer mineralized zone, were observed. No reason for

the blackness of the intermediate zone is known, but some day the answer

may be given by electron beam microprobe analysis or by a similar tech-

nique.

The outer mineralized zone or the corrosion crust has a layered structure

made up mainly of elemental (redeposited) copper, oxides and salts of

copper, and oxide of tin, which will be described in some detail in Chapter

VIII. Studies on these three zones confirm the view already expressed by

Fink and Poluskin 153 and by Caley 154 that the mechanism of corrosion

varies considerably from piece to piece and even from place to place within

a single vessel. This results from factors in the local environment, such as

moisture content of the soil, salts, organic matter, bacterial action, per-

meability to atmospheric oxygen and contact with other metals. The results

cannot be predicted from short-time laboratory tests. In general, attack

probably occurs by a slow capillary mechanism which results first in the

destruction of lead particles along grain boundaries, then of the tin-rich

portions of the eutectoid and finally of the more copper-rich skeletons of

the dendrites. Corrosion must always proceed continuously with access

to the outside environment. The mineral that is formed first completely

fills up the space formerly occupied by the removed metal, and so preserves a

record of its shape. The arrangement and sequence of the residual products

as they most commonly appear in deeply corroded bronzes suggests that

a localized electrolytic process causes the a bronze to be replaced with

porous metallic copper mixed with tin oxide while the cavities left by

shrinkage or by the removal of lead provide a place for the deposition of

elemental copper. In time (beginning at the outer parts of the casting where

153" Microscopic study . ..," p. 114.

154" corroded bronze of Corinth."

130

more oxygen is available), the copper is converted to relatively solid cuprite

(still with tin oxide mechanically incorporated in it) and finally the cuprite

reacts with carbonic acid to form malachite, which is a stable mineral and

undergoes no further change. Because the corrosion-product minerals

occupy considerably more space than the metal from which they form,

some of the copper ions that are formed at the metal interface diffuse or

flow in solution through capillary channels until they find space outside

the original surface of the bronze where they can form crystalline mala-

chite. This is fortunate, for this external layer is relatively soft and friable

and can be removed to reveal the original surface of the bronze. (With iron,

however, the corrosion products expand in situ with sufficient force to

destroy the surface completely.) The composition and physical nature of

the more massive corrosion products will be discussed in Chapter VIII.

Special studies on a chaplet

Although as described in Chapter IV some of the chaplets in vessels are

plainly visible, it was difficult to get much specific information about com-position, physical structure, and relationship to the matrix metal from direct

observations. By good fortune, a broken p'an {fig. 163) came into the study

collection; it was possible to break out from the bottom a small fragment

of metal with a chaplet in place and to mount and polish it to expose the

chaplet in situ. This and other chaplets are well shown in the X-ray of the

p'an (fig. 164). Figure 165 shows the whole section after polishing and

etching.155 There is no sign of fusing of the chaplet with the matrix metal;

but since the edge of the chaplet is concave on one side and irregular on the

other, it is well locked in place mechanically. A narrow zone of the matrix

metal on either side of the chaplet is different in appearance from metal

farther removed from the contact zone. Figure 166, at higher magnification,

shows that this zone is less heavily corroded than the bulk of the metal, andapparently corresponds to a chilled zone in which the molten metal wasrapidly cooled by coming into contact with a relatively cold chaplet. Duringsolidification the crystallizing metal has rejected lead, and there is a faintly

enriched zone marking the termination of the chill. It can be noted that the

chill is much greater on one side of the chaplet than on the other, perhaps

as a result of the direction of the initial flow of metal.

The chaplet is cast bronze. Its appearance suggests that it might have been

slightly flattened mechanically, but the structure of the metal shows noevidence of deformation, and the interlock is purely that of molten metal155The metallographic study was made by Professor Cyril S. Smith at the Massachusetts Institute of Tech-nology, and the metallographs were made by Mrs. Katharine Ruhl. Much of the information given here is

paraphrased or used directly from Smith's report privately communicated (24 November, 1965).

131

Figure 165

Figure 166

Figures 165-168. P'an SC537. A section through a chaplet in the vessel bottom. Figure 165. The chaplet, of

cast bronze, is firmly gripped by the metal which was cast around it ( 8, No. 080-A). Figure 166. The metal-

lograph covers the junction of the chaplet (left) with the main casting (right). The light-etching area seems

to correspond to metal that solidified rapidly as a result of the chilling effect of the chaplet ( x 50, No. 080-A7).

Figure 167. Inter- and intragranular corrosion is plainly seen in the chilled zone of the matrix metal. Thecorrosion product is mainly cuprite ( x 200, No. 080-H3). Figure 168. The microstructure of the matrix metal

at higher magnification shows various kinds of mineral products resulting from corrosion. The bright areas

are uncorroded metal; the gray areas are cuprite; the darker areas are tin oxide from S phase corrosion; the

boundaries and slip lines that begin as tin oxide widen into cuprite (x500, No. 080-A9). All metallographs

were taken with bright field illumination after the section was etched with ammonia and hydrogen peroxide.

(Courtesy, Katharine Ruhl, Massachusetts Institute of Technology.)

132

running against the shaped surface and solidifying without actually melting

it. The grain size of the metal in both the casting and in the chaplet is quite

small which supports the belief that the metal had been cast in a relatively

cold mold. The grain size is slightly larger in the chilled areas for here the

grains were growing unidirectionally without interference. The metal in the

body of the casting would have been in a turbulent state with many crystals

growing in it.

Most of the visible structure is mineral corrosion product. The local

difference in the extent of corrosion is mostly a result of the change of

structure due to the chilling operation. Grain boundary corrosion is less in

the chilled area than elsewhere, though grain boundaries have clearly pro-

vided a route for corrosive attack on both the eutectoid 8 phase and the

lead. In a few parts of the chilled area, the corrosion in the a phase tends to

follow crystallographic directions (probably on octahedral planes)—

a

behavior which often seems to follow slight deformation of the metal prior

to corrosion. This is seen in figures 167 and 168. The most important fruit

of the examination is the evidence that the chaplet was in place before the

main body of metal was run into the mold, and that the mold was not very

hot.

Metallographie study ofsolder in the handlejoin o/kuei 24.11

Laboratory examination revealed that the vessel is of unusual interest

because both handles are joined to the bowl with a ribbon of hard solder

in a way and under circumstances that permit chemical and metallographie

study of the solder alloy (see fig. 169; also Chapter IV, section 12). Afew other vessels have hard solder joins, but on those there was only limited

opportunity for technical study. On this kuei the lower part of one handle

was separated from the body because the solder apparently had not wet

the body metal sufficiently to make a proper join; the top part, however, was

tight and secure indicating that here the solder had wet and adhered to the

contiguous surfaces. Because the vessel was only useful as a study piece, 156

the lower portion of the handle was cut away so that the metal-solder rela-

tionship could be conveniently studied. As shown in figure 169, the solder,

which even to the naked eye shows a heterogenous structure, clung to but

did not wet the handle metal at the point of the saw cut. Higher on the same

join there is no doubt of better contact because after many centuries the join

is still secure.

156This kuei was purchased in 1924 by the late Carl Whiting Bishop when he was in charge of the Freer

Gallery Field Expedition to China. At that time, he bought a good many objects of various kinds for study.

Years later, some time in the 1950's, Mr. Wenley added a few of these pieces to the collection giving them 1924

accession numbers retroactively.

134

Figure 170

Figure 171

Figures 169-172. Kuei 24.11. This deeply corroded vessel

(see PI. 81, Vol. I) had apparently come directly from

excavations and had not been "restored." Its condition madeit suitable for technical studies. Figure 169. A section cut

from the underside of the handle (see figs. 79 and 80, Ch.

IV) reveals an irregular mass of heterogeneous hard solder

used to join the member to the vessel side ( x 2.5). Figures

170 (No. 025B-2) and 171 (No. 025B-7). The general den-

dritic structure of the handle metal is evident at low magni-

fication (x200, fig. 170) after etching with potassium

dichromate. The view at higher magnification ( K 1000,

fig. 171) shows patches of eutectoid (hatched areas) which

result from the decomposition during cooling of the y

Figure 172

phase which breaks up into a and S below 520°C. There are

also rounded pools of lead (gray), sometimes replaced by a

cavity (black) where shrinkage has drained away the lead.

There are also a number of non-metallic inclusions (light

gray) presumably of cuprite. Although the etchant did not

accentuate the composition gradients in the a phase,

evidence of a cored structure is revealed by the mottled

appearance of the lighter areas. Figure 172. Microstructure

of bronze near the edge of the handle; etched with potas-

sium dichromate ( x 500). Penetration of corrosion prod-

ucts along crystallographic planes suggests that the metal

was to a degree cold-worked. (Courtesy, Katharine Ruhl,

Massachusetts Institute of Technology.)

135

Figure 176

Figure 177

Figures 173-177. Kuei 24.11. The solder lump attached to

the end of the handle {fig. 169) was pried away for metallo-

graphic examination. Figure 173. A cross-section of the

solder lump, etched and lightly re-polished (x7.5). Thelighter parts are uncorroded bronze; the dark areas are

those in which the S phase has been entirely converted to

corrosion product. The circular white parts in the corroded

background are electrolytically redeposited copper occupy-

ing the space of the lead particles which can be seen as

black circles in the uncorroded metal (No. 025A-1).

Figure 174. Microstructure of the solder after etching with

potassium dichromate. Here the similarity in shape be-

tween lead droplets and redeposited copper is quite ap-

parent. Straight parallel lines in the copper globules in-

dicate the presence of twinned crystals (No. 025A-5).

Figure 175. (No. 025A-15). Microstructure at the edge of

the corroded areas of the solder, unetched ( x 1000). Thecavity (black area) is left by the dissolving away of a lead

particle in the corrosion process. In more extensively cor-

roded areas such cavities are filled by deposition of copper

(see fig. 174 above). The dark mottled areas are eutectoid

in which the 8 phase has been converted to tin oxide, the a

remains unaltered. Figure 176 (No. 025A-14). The samearea after etching with potassium dichromate (xlOOO).

This reveals the structure of the uncorroded eutectoid (gray

areas in uncorroded portion above). Figure 177. Micro-

structure of uncorroded solder; etched with potassium

dichromate ( x 1000). (Courtesy, Katharine Ruhl, Massa-

chusetts Institute of Technology.)

137

Chemical analysis gives the following results

:

TABLE XI

Analysis of kuei 24. 1

1

Cu Sn Pb Total* Estimated

% % % % melting point

Metal of vessel 65.8 7.2 22.8 95.8 900 °CMetal of handle 76.4 8.0 12.4 96.8 940

Solder: sample a 60.4 22.9 9.4 92.7 800

Solder: sample 6 64.5 19.2 9.2 92.9 810

*Low totals are mostly caused by the presence of corrosion products, largely cuprite.

After preliminary study of the sawed-off section, further small samples of

both handle metal and of solder were detached, and these were mounted in

plastic, polished, and etched in preparation for examination of the micro-

structure at high magnification. 157 The handle metal is reasonably sound, andthere is only a thin layer of corrosion product on the surface. The micro-

structure is quite typical of a cast alloy of the composition shown in the

analysis (Table XI). The general dendritic appearance is evident at low

magnification {fig. 170). Higher magnification {fig. 171) reveals rounded

patches of lead and patches of sharply outlined eutectoid. The lead is

sometimes replaced by cavities where the shrinkage of the liquid metal or

perhaps gas evolution has drained away the lead during the final stages of

solidification. The islands of eutectoid show a fine structure which results

from decomposition during cooling of the y phase. The structure is com-pletely normal for a cast metal not subjected to any mechanical or thermal

treatment after solidification. A partly cold-worked structure can be seen in

an area near the outside surface {fig. 172).

The solder is a very different alloy (Table XI). It is quite apparent from

the general view (fig. 173) that it had not been applied in the modern mannerof brazing, and in this particular part of the handle had not properly wet

(i.e. metallurgically joined to) either the bronze vessel or the handle. Thegeneral appearance of the join in this and in other vessels suggests that the

solder was applied as a blob of metal in an incompletely melted pasty con-

dition, and the handle pressed into it. The microstructure of the solder

supports this supposition (figs. 174-177). In the uncorroded parts most of the

field consists of a background of roughly equal quantities of small crystals

of unconverted a phase and of eutectoid. The eutectoid is quite finely divided

which indicates moderately rapid cooling, perhaps by contact with the

157The observations reported here were also made by Professor Cyril S. Smith, privately communicated(8 December, 1964) and the metallographs were made by Mrs. Katharine Ruhl.

138

cold metal of vessel and handle; but there are irregularly distributed, more

or less spherical, blobs of lead, the largest of which measures 170 microns in

diameter and is just visible to the naked eye. The a crystals have no visible

concentration gradient in them and a small fraction, perhaps 5 percent, are

twinned. Unstrained cast metal does not contain twins and their presence

indicates that the metal had been deformed after the a phase had already

crystallized.

The solder is much more corroded than the handle metal, probably be-

cause the patches of 8 plus eutectoid in the latter are isolated while in the

solder they are contiguous and provide a connected path for corroding

agencies.

The use of copper-tin solder in a pasty condition is a high temperature

analogy to the modern plumber's wiped join in which a 50-50 lead-tin alloy

is used in a pasty condition. It would be quite effective if the surfaces to be

joined were hot enough to be wet by the liquid phase in the paste.

These three case histories show that much can be learned from micro-

scopic examination of properly prepared samples of ancient bronzes.

Recognition of various intermetallic phases permits us to judge the approxi-

mate copper-tin ratio. The lead content can also be estimated from the size

and distribution of the lead globules. Size of the crystals of a solid solution

give a clue to the rate of pouring of the casting and the rate of cooling of the

melt. Cold working after casting can be recognized; and finally the extent

and nature of corrosion, both surface and intergranular, can be seen.

139

CHAPTER VI

INSCRIPTIONS

One of the more puzzling features of early bronze vessels is the mannerin which the inscriptions were produced. On most vessels the characters are

in intaglio, although occasionally they are cast in relief. In some of the

Shang inscriptions the characters have raised edges as if the strokes had

originally been pressed or drawn with a stylus in some plastic material

{figs. 178-181). On the other hand we do not yet have sufficient evidence

to say that these raised edges are characteristic of inscriptions of the period.

The characters of the much longer inscriptions on many of the Choubronzes seem, however, to lack the plastic quality just described. If weexamine closely the characters of a huo 33.2 {fig. 182), we can see that the

edges are flat, not raised; the sides of the lines are wavy and in some in-

stances the vertical sides of the strokes are actually undercut. The wavysides of the characters in a fang-i 30.54 {fig. 183) are typical. The grooves of

these characters do not appear to have been drawn originally in a plastic

medium, and they do not look as if they were engraved with a tool after

casting. At best we can only conjecture as to how they were made.

Noel Barnard seems to be the only Westerner who has grappled with the

problem of making inscriptions, 158 and in his opinion it was part of the main-

casting operation. He suggests various methods that might have been

employed, but holds strongly to the view that the inscription was cut or

engraved first on a clay block in intaglio ; then a corresponding small block or

slab of clay was pressed on the surface of the intaglio; and this in turn gave,

on separation of the two blocks, characters in relief. This clay block was

then inserted in a prepared recess in the bottom of the core and later, whenthe bronze was cast, intaglio characters resulted. Barnard shows diagrams

to illustrate how he thinks this was done.

Inscribed bronzes in the Freer and other collections provide some sup-

port for Barnard's theory. For example, the single character inside the cover

of the lid of a chih 38.6 is set in a shallow depression bordered by the almost

imperceptible sunken lines of a square {fig. 184). The single character inside

the foot of another ku 07.34 {fig. 185) is even more deeply recessed. On the

other hand the single character in the bottom of chia 35.12 is slightly raised

and is framed with faint sunken lines barely distinguishable in strong raking158Bronze casting . . . , pp. 157-168.

141

Figure 178

Kit 40.3. The edges of this two-character inscription are raised as if the character originally was

stamped in a plastic substance. Note also that the bottoms of the channels are fiat (macro-

graph x 7).

Figure 179

Ku 43.9. The three-character inscription on the inside of another ku also has raised edges; the

cuprite which lines the channels has been partially removed by mechanical cleaning (macro-

graph x 6.5).

142

Figure 180

Ku 51.18. The two-character inscription on the inside of the foot of another Shang dynasty ku

has raised edges and flat-bottom channels (macrograph < 7).

143

Figure 182

Huo 33.2. This inscription on a Chou dynasty vessel has flat, nearly square edges and the

channels are rounded. In some places the vertical walls of the channels are slightly undercut

(macrograph 5.3).

Figure 183

Fang i 30.54. The edges of the characters in the lid are wavy but flush with the surface which

seems typical ofmany Chou inscriptions (macrograph x 2.7).

Figure 184

Chih 38.6. The single character in the lid is set in a shallow

square depression faintly outlined by sunken lines

(arrows). These lines may mark the edges of a block set

in the core to carry the negative mold (raised lines) of the

character.

Figure 187

Yu 60.20. A common feature of inscriptions of the Chouperiod is the undercutting of the edges. The undercutting

of a character in an inscription in the lid of this vessel is

so deep that the wire loop is partially concealed by the

overhang (macrograph x 8.5).

Figure 185

Ku 07.34. The single character in the inside of the foot is

set in a low rectangular depression which indicates that

the negative mold for the character was made separately

and fitted into the core (macrograph X 1 .2).

Chia 35.12. The single character in the bottom is slightly

raised and framed in a rectangle of faint sunken lines

(macrograph x0.9).

Figure 188

A character recently etched in a cast blank of bronze.

Dilute nitric acid was used as the mordant and wax-

resin as the resist. Note the rounding of corners, the

jagged sharpness of edges, and the roughened surface at

the bottom of the groove (macrograph x 3).

light {fig. 186). Karlbeck has made the same observation on bronzes in other

collections. He says (private communication) that in some cases there is

just one line bordering the inscription and that it may only be visible in a

raking light. Karlbeck agrees essentially with Barnard that the inscription

mold was prepared separately and that it was set as a block into the core.

None of these suggested methods, however, seems to be entirely satisfac-

tory. In many instances the strokes of the individual characters do not look

as if they had been carved or engraved in clay or other medium; nor do they

look as if they had first been built up in clay impasto. There is no suggestion

from their line quality that the characters were cast from an intaglio moldalthough we must admit the characters as we see them would be two steps

removed from the original inscribing. Second, if blocks bearing an inscrip-

tion had been inserted in a recess in the bottom of a core as Barnard de-

picts,159 it seems they would have left evidence of a join or mold mark at

the boundaries of the insert even more frequently than they do. And, third,

among all the mold fragments so far discovered, there is nothing to support

any such theory of inscription making, although it must be admitted that

perhaps the molds were so completely mutilated and destroyed after casting

that it would not be possible to find mold inserts or sections with inscrip-

tions if they had existed.160 The casting of inscriptions in relief would seem

easier than in intaglio, but this may be one more instance of the uniqueness

of Chinese technology.

When the writer first examined these early Chinese characters some years

ago, he gained the impression that they were formed in the metal by acid

etching. They looked that way because the edges of the characters of manyinscriptions are undercut which is a common characteristic of etched lines

in metal; also as mentioned before the edges of the lines are often wavy.

The undercut feature is well illustrated by characters of the lid of a yu 60.20

which seem to lack either drawn or plastic quality (fig. 187). Undercutting,

however, is not restricted to inscriptions but can also be observed on the

edges of sunken decor in late Chou bronzes (see Chapter IV).

The idea of acid etching struck the writer so forcibly that he conducted

some crude experiments to see if such characters could actually be cut in

copper alloys by acid. A few scrap pieces of bronze were coated with wax-

resin mixture and, when cooled, Chinese characters were drawn in the waxwith a wooden stylus deep enough to lay bare the metal. The metal was then

etched with nitric acid diluted 1 :2.5. Nitric acid was chosen for its conve-

nience and rapid action. It was possible in this way to get intaglio characters,

1590/?.c/7.,p. 160.160Barnard says (private communication) that two inscribed core fragments have recently come to light butunfortunately they are illustrated only by rubbings.

146

deeply cut with sharp but wavy edges (fig. 188). It was also easy to get

undercutting if the etching was prolonged. The entire idea of etched in-

scriptions, however, was soon dismissed, because there seems not to be a

shred of outside evidence, literary, historical or technical to support it.

Later on, however, the writer was impressed when three other persons, two

of them metal experts who had had little experience with Chinese bronzes,

separately and independently examined our early inscribed vessels and

likewise observed that the characters "looked acid-etched. " This forced

reconsideration of the discarded theory of etching. There was one formid-

able obstacle, however, in the way of developing a theory of etching as a

means of writing inscriptions, and that was thequestionofwhetherthe know-

ledge of strong acids existed in Chou times and whether any suitable acid

or etching mixture was then available to do the job. It is known that in Hantimes the Chinese were versed in the chemical arts, and there is reason to

believe that the black art of alchemy was also practiced. 161 Any culture

which was as accomplished as that of the Chinese in metallurgy, a special

branch of chemistry, must have had some knowledge of acids, bases and

salts. Certainly saltpeter, KNO3, which is the starting point for nitric acid

was known to the inhabitants of any Asiatic city located in a semi-arid

climate. But we do not mean to imply that we think the ancient Chinese

made bronze inscriptions with nitric acid. Our purpose is to explore all

the possibilities of making lines in metal so that when we examine morebronzes our vision will have greater acuity. It would be possible to conjec-

ture at length on the theme of etching and even attempt to relate backwards

to ancient bronzes the late medieval formulas for etching armor and dama-scening swords, but that would seem to be a digression at this time.

The authenticity of the inscriptions found in bronze vessels is sometimes

questioned on grounds of epigraphic style, historical content or demon-strable deviation from more safely attested texts. It is sometimes inferred

that they have been incised in recent times in bronzes which themselves are

truly archaic. There are a few real examples of this. Plenderleith described

with illustrations an inscription which was chiseled into a bronze vessel.162

The hardness of the line, and the cracking of the brittle surface caused by

the tool are quite obvious. Plenderleith cites another example (p. 53) of

an inscription incised into the neck of a vessel and concealed with false

patina. After removal of the false patina, chisel marks in the bottom of the

grooves of the characters were easily visible. Rudolph described an altered

161 See Masumi Chickashige, Oriental alchemy; Obed S. Johnson, A study of Chinese alchemy; Li Ch'iao-ping,The chemical arts of old China; also the writings of Joseph Needham of Cambridge and of the late ProfessorTenney L. Davis of Massachusetts Institute of Technology.162"Technical notes on Chinese bronzes . .

.," p. 52.

147

Figure 189

Figure 190

Figures 189 and 190. An 11.45. View of underside. Figure 189. The inscription, strangely located on the

underside of this vessel, is a forgery. Figure 190. It appears to have been cut or chiseled into the cor-

rosion layers. The edges of the top layer, which is cerussite, are frayed and chipped, especially on the

tops of the little islands that form the scales of the fish (macrograph x 9).

148

inscription on the side of a bronze bell in the Royal Ontario Museum.163

Here one part of the original cast inscription was removed and replaced by

court officials of China in the year 1174 because it violated the taboo

relating to the use of a personal name of a former Chin ruler. The raised

edges of the engraved inscription contrast strongly with the flush and smooth

edges of an unaltered inscription on the opposite side.

Doubt has been cast on some of the Freer inscriptions and the reasons

for reservations are mentioned elsewhere; but all have been examined

carefully in the laboratory in good light at some magnification, and only a

small number can be questioned on purely technical grounds. An obviously

incised inscription occurs on the underside of an 11.45 {fig. 189). The

two characters, one in the form of a fish, are cut deeply into the hard com-

pact corrosion layers of the surface, and it can be seen in the photomacro-

graph {fig. 190) that the strokes have an incised look, not a plastic one.

This inscription is unique in another way; it is the only one in the entire

collection which is located on the underside of a vessel, the only place where

a forger could work freely. The hand cutting of an inscription inside the

bottom of a deep vessel would seem to offer some difficulties, but when weconsider the skill of the Chinese jade carvers perhaps it is not impossible.

There is another curious aspect of the making of inscriptions. It was

mentioned in Chapter IV, section 21, that the three chaplets in the bottom

of kuei 11.38 are actually crossed by strokes of some of the characters.

The authenticity of this inscription has been doubted by Noel Barnard on

textual grounds and in fact he calls it "spuriously incised." Careful examin-

ation under magnification shows that the characters are made in the normal

manner of Chou inscriptions {fig. 191). There is no indication of incising,

cutting or chiseling in recent times. It happens that the underside of this vessel

bears the kind of criss-cross lines mentioned in Chapter IV, section 6, and a

radiograph {fig. 192) shows that these lines actually extend over the edges

of the three chaplets. Figures 193 and 194 show in detail the front and rear

of one of the chaplets. This circumstance adds to the already overwhelming

evidence that the chaplets were in position at the time the vessel was cast. It

raises, however, a rather awkward question: how can a cast-in character

cross over a chaplet ? On the other hand if the stroke is incised why is it

interrupted in the middle of the chaplet? In another chaplet of the samevessel crossed by a character (see Chapter IV, fig. 125) the strokes are not

interrupted. The point is that a cast-in character should stop at the edge of

the chaplet, whereas an "incised" character should cross over the chaplet

without interruption.

163"Dynastic booty "

149

Figure 192

Figures 191 and 192. Kuei 11.38. Many of the Chou dynasty vessels have long inscriptions on the

inside bottom. Figure 191. Three characters of the inscription, indicated by arrows, touch or cut across

edges of chaplets; otherwise the inscription seems normal for inscribed bronzes of this period. Figure

192. The radiograph of the bottom clearly shows the inscription, the three chaplets, and in the lower por-

tion, one chaplet which is crossed on the front side by strokes of a character and on back by raised

criss-cross lines.

150

Figure 194

Figures 193 and 194. Kuei 11.38. Figure 193. The character wang which crosses one of the chaplets is

incomplete having part of the vertical stroke missing. Note that the adjacent character pao which is

not involved with a chaplet also has an interrupted stroke. Figure 194. Underneath the vessel the chap-

let shows equally plainly, and the edges are partially crossed by the raised criss-cross lines that are com-mon to many vessels of this period (macrograph x 5).

151

152

153

This is not an isolated example. The lines of the 12 character inscription

of the fragmentary p'an SC537 {fig. 195) already mentioned, cross over someof the many chaplets in its bottom. The radiograph (figs. 196 and 197)

shows the relationship especially well. Since the object was expendable, a

section of metal about 1.5 cm. square was cut out in a location where a

character actually crosses a chaplet, and in such orientation that whenmounted for grinding and polishing the intaglio lines would be at right

angles to the direction of polish. For control purposes a similar section was

cut from another character not crossed by a chaplet (fig. 198). Mountedand polished in the usual way for metallographic study and photomicro-

graphy, these sections revealed some interesting facts. The nature of the in-

taglio lines is clearly shown in the macrographs (figs. 199 and 200). The

most important single observation is that the groove of the stroke which

cuts across the chaplet is exactly the same as those which cut across the

surface of the vessel proper. Obviously they were formed by the same me-

thod. Secondly, in all five grooves including the one that crosses the chaplet,

the cut on one side of the groove is nearly straight down while on the op-

posite side it is slanting, giving a sort of saw-tooth effect. The same saw-

tooth profile exists in a third character of this inscription. This was demon-strated when cross-sections were made of a silicone rubber impression of the

character (fig. 201). (This technique does not require sectioning of the

bronze.) The characters were perhaps cut with an abrasive tool in muchthe same way that lines in jade were cut. There are no slip lines in the un-

corroded metal grains or other indication that the lines were cut by sharp

blows applied with a chisel. The broad surface and low rim of the vessel

would make such an operation possible. A third but equally important

observation is that the metal throughout is deeply corroded and that

corrosion extends inward unaltered and undisturbed from the edges of

the grooves (figs. 202 and 203). Between a third and half of the original

metal has been replaced by cuprite. It is further of interest to note that the

walls of each one of the grooves is lined with a thin and nearly continuous

layer of cuprite. Most important, there is no indication that the grooves

were cut orformed subsequent to corrosion. The inscription, however it was

made, was placed in the bottom of the vessel prior to corrosion, and

presumably prior to burial.

The inscriptions, like the vessels themselves, were probably made by

varying methods. It is a marvel that such an important facet of vessel

casting technology should so long defy rational explanation. From the

foregoing experiments, probings and observations it is clear that much

154

work must still be done before the final word on the techniques of inscribing

ancient bronzes can be written.

155

Figure 201

Figures 198-203. P'an SC537. (See figs. 195-197.) Figure

198. The detail (macrograph x 3) shows where two sec-

tions, selected for metallographic purposes, were cut from

the area of the inscription. The lower section cuts across a

chaplet which in turn is crossed by the strokes of a charac-

ter. The sections were mounted edgewise and polished. Theupper section (fig. 199) cuts across three strokes of the

second character in the left line. The lower (fig. 200) cuts

across two strokes of the character below it, which in turn

cut across a chaplet. The right sides of the strokes are cut

steeper than the left (macrograph x6.3). Figure 201.

Selected cross sections of a silicone rubber impression from

the last character in the right hand column of the inscrip-

tion show the saw-tooth profile of the grooves, their rela-

tive depths, and the gradual slope at the ends of the lines

(macrograph x 5). Figure 202 shows the metal structure in

the region of one of the strokes. Figure 203 shows where

a stroke cuts across a chaplet. The walls of the strokes are

thinly lined with cuprite and much intergranular cuprite

permeates the whole metal structure indicating that cor-

rosion of the bronze was subsequent to the making of the

inscription (bright field X 50).

157

CHAPTER VII

RADIOGRAPHY

It has long been recognized that X-rays and other rays of high energy are

useful in the examination of metal antiquities, but they have not been

employed so widely for this purpose as for the study of paintings because

of the greater opacity of metals to X-rays and the higher energies and longer

exposures required. For bronze vessels with wall thicknesses ranging

between 1 mm. and 1 cm. and alloys with lead content sometimes as high

as 25 percent, it is necessary to employ kilovoltages considerably higher

than those supplied by the standard medical units that are ordinarily

available in museum laboratories.

Following preliminary X-ray studies on miscellaneous bronze objects in

the Freer collection, officials of the U.S. Naval Ordnance Laboratory at

White Oak, Silver Spring, Maryland, agreed to undertake the radiography

of a limited number of our bronze vessels principally to try to learn moreabout structures, methods of fabrication and joining techniques and second-

ly, to reveal the kind and extent of repairs, especially modern repairs madewith tin-lead solder. A total of 28 vessels were X-rayed, 25 from the Freer

collection and 3 from the study collection. The radiographs produced are

highly informative. A number of them have been used to illustrate matters

touched on in Chapter IV and a few are brought together here to show their

general applicability in bronze studies.164

Problems ofstructure

The radiographs reveal clearly the clay cores that are so often present in

legs, handles and decorative appendages. The thin walls of the cored parts

are clearly and precisely revealed. It was known that the knob on the lid

of the large yu 30.26 was clay cored, but the radiograph revealed in addition

that the core is conically shaped (see Chapter IV, fig. 107). It also showedthat the inside surfaces of the two cored lugs of the lid are nearly as finely

finished as the outside (figs. 204 and 205). The radiographs have also

164The majority of the radiographs were taken with a 250 KVP industrial X-ray unit ranging from 120 KVPto 250 KVP at ten milliamperes, at distance four to five feet (100-125 cm.) and exposure time 40 seconds to six

minutes. Where greater penetration was needed, radiographs were taken with a 2000 KVP X-ray unit using2 Mev. 1.5 ma., exposure time 15 to 30 seconds at distance of 9 feet (225 cm.). All radiographs were made onKodak Type M industrial X-ray film which in most instances was developed in an automatic X-ray processor(Kodak X-omat); in order to achieve better contrast, a few films were developed by hand in Microdol ac-

cording to standard procedure.

159

Figure 205

Figure 204

Figures 204 and 205. Yu 30.26. Figure 204 shows the heavily constructed lid of the vessel. Figure 205. Theradiograph shows that the two animal-head hand grips and the knob have clay cores; and also that the inside

walls in contact with the cores are regularly and precisely formed.

Figure 206 Figure 207

Figures 206 and 207. Yu 30.26. Figure 206. View looking down on one of the four appendages that are

joined to the body with hard solder. Figure 207. The radiograph shows that the long earthen cored neck is

fitted over a boss which apparently was cast integrally with the vessel.

160

Figure 209 Figure 210

Figures 208-210. Fang-i 30.54. Figure 208. The knob and side-view radiograph shows the stem core is not centered,

stem on top of the lid are cast integrally. The radiographs Figure 210. The light area in the end view of the knob in-

show that the knob and stem are cored. Figure 209. The dicates that some metal ran into a hole in the core.

161

cleared up some questions about how the four long-necked animal head

appendages of the vessel were joined to the body; the hollow end of each

appendage was fitted over a boss precast into the side to provide reinforce-

ment for a hard-solder join (figs. 206 and 207). It was surprising to learn

from a radiograph of the lid of our famous fang-i 30.54 that the hipped

knob and its stem are cored, and furthermore that the core of the stem is

tilted and off center (figs. 208-210). Radiographs of the bird-capped posts

of chia 07.37 mentioned in Chapter IV show that the birds are cored and

cast on to conical caps on the top of the posts (figs. 211 and 212). The bosses

which serve to secure the cored animal heads on the side of tsun 51.19 are

clearly shown (see Chapter l\,fig. 109) but the exact method of attachment

to the bosses remains unknown. Radiographs of legs and handles secured

to vessels by interlock casting fail to show the exact location of the seam of

the join.

It was known that the handles of kuei 38.20 were precast, and the vessel

cast to them. In the radiograph the handles are shown to have square ends,

which do not abut the vessel precisely hence it is remarkable that they are

so firmly secured by the thin metal of the vessel sidewall that surrounds

them (fig. 213).

Ancient repairs

Radiographs are often helpful in revealing ancient repairs. It was knownthat three of the legs of afang-ting 50.7 are solid and integrally cast with the

vessel, but that the fourth leg, which is clay cored, is a later addition. This

was confirmed by X-ray (fig. 214) which also shows that the core of this

leg is improperly aligned. The radiograph, however, does not show any

details of the interlock join between vessel and leg. A repair area of plaster

and paint on the surface of a p'ou 09.334 raised many questions about con-

dition, but a series of radiographs confirmed earlier judgement that there are

no losses of metal or serious breaks in the side wall and that the reason for

the repairs was simply that deep corrosion in this area had created a pow-

dery surface which required faking of decor (fig. 215). On many vessels

radiographs clearly reveal the location of chaplets when they are well

concealed by metal overflow, by fine finishing, or even by corrosion (figs.

216 and 217). A radiograph of the bottom of a large chien 15.107 not only

reveals a number of chaplets randomly placed, but also some evidence that

at the time of casting the inflowing molten metal partially melted the edges

of the chaplets (figs. 218-220). Blow holes and other casting flaws are also

plainly revealed (fig. 221). The decor on the surface of many vessels is some-

times shown in finer detail and better contrast in radiographs than in

162

Figure 21

1

Figures 211 and 212. Chia 07.37. Figure 211. One of the two birds surmounting the rim of the vessel. Theyare core cast. Figure 212. The radiograph shows that they are fixed to solid posts which terminate in conically

shaped caps.

Figure 213

Kuei 38.20. The radiograph shows clearly that the handles

are cored and precast and that they are held on by metal of

the vessel that surrounds them (arrows).

163

Figure 214

Fang-ting 50.7. The radiograph shows that three of the legs are solid and cast integrally with the vessel, but a

fourth leg which is a later addition or repair is clay cored and cast on to the vessel (see also figs. 139 and 140,

Ch. IV).

164

Figure 215

P'ou 09.334. An irregular area of the side wall of the vessel is covered with paint and plaster repair, but the

radiograph shows there is no break or loss of metal, only extensive corrosion which necessitated the repair. Asegment of the cast side wall on the left is noticeably thinner than the one on the right.

Figure 2 1 6 Figure 2 1

7

Figures 216 and 217. Kuang 49.10. Figure 216. Top view of the lid. Figure 217. The radiograph reveals the

presence of a chaplet directly above each eye of the tao tieh mask and one behind each ear (arrows). The first

two are visible on the surface; the others are concealed by metal overflow and corrosion.

165

Figure 218

Figure 219

Figures 218-220. Chien 15.107. Figure 218. A view of the

bottom shows a thin ridge across the centre which appears

to be the stump of a sprue. Figure 219. The all-over radio-

graph shows a number of chaplets rather randomly distri-

Figure 220

buted over the bottom. Figure 220. The halo-like zone

about each chaplet, shown in an enlarged detail, indicates

that the heat of the liquid metal had started to fuse their

edges.

166

167

Figure 223

Figure 224

Figures 223-225. Ting 47.20. Both vessel and lid have been

extensively repaired. Figure 223. The lid has various decor

and repair features better revealed in the radiographs than

in the photograph. Figure 224. A break in the repeat pat-

tern of hooks and volutes on the lid is clearly shown(arrows) in the detail radiograph. The white rectangular

Figure 225

area is a handle. The jagged white streak is a soft-solder

repair. Figure 225. The all-over radiograph shows that the

lid has been crushed in but it was possible to reassemble the

fragments and join them with lead-tin solder (white lines)

and to conceal joins with imitation patina.

Figures 226 and 227. Hu 49.5. The appearance in ultra-

violet light indicates that the vessel has been "restored."

Figure 226. The radiograph reveals that the vessel has been

damaged and extensively repaired. One large area of loss

has been replaced with an insert made of lead. Figure 227.

As seen in the detail, the surface of the large lead repair in-

sert in the side of the vessel has been carved and modeledto match the surrounding sunken decor. The repair area is

concealed with a kind of paint that imitates the natural

patina. It is the paint that causes the area to stand out in

ultraviolet light.

169

photographs (fig. 222). They also sometimes separate remarkably well the

levels of sunken decor as is shown in a radiograph of the lid of ting 47.20,

where the joins between abutting decor units are plainly visible {figs. 223

and 224).

Problems ofcondition andmodem repairs

It was recognized that a number of the vessels had been repaired in

modern times, because these repairs were usually made with tin-lead solder,

and the joins and replacements concealed with paint and artificial patina.

X-rays are especially useful in locating soft solder joins because of the high

absorption (opaqueness) of lead to X-rays; on X-ray films the solder joins

show as transparencies. The radiograph of the lid of ting 47.20 shows that

the center of the lid was crushed and broken into six fragments which were

all put back into exact position and joined with soft solder, and the join

concealed with false patina (figs. 223 and 225). The extent of the break and

repairs in the edge of kuang 49.10 is clearly shown by the radiograph (see

Chapter X, fig. 302). A broken-out portion of the side wall of hu 49.5

(fig. 227) has been repaired with a lead insert modeled to match the

surrounding decor; a radiograph (fig. 226) shows the lead replacement and

the great extent of damage in the area.

These studies suggest that the condition of many bronze vessels could be

better revealed by X-radiography. X-rays are as important in bronze

connoisseurship as in the study of ancient paintings. The only barrier to

their wide use in examination of metal antiquities is the lesser availability

of the higher energy industrial X-ray units required to get satisfactory

radiographs.

170

CHAPTER VIII

PATINA AND CORROSION

Often the collectors' interest in ancient bronzes is quickened by the

colored corrosion products or patina of their surfaces. The term patina,

derived from the Latin patina (a dish), was used originally to describe the

green, blue and red corrosion products of copper and bronze, but by exten-

sion is now used to describe any surface on wood, ivory, or marble or other

materials resulting from age and exposure. Chemists use the terms "cor-

rosion" and "corrosion products" to describe the salts and oxides that form

during burial or long atmospheric exposure, while mineralogists prefer to

call them "mineral alteration products." Patina, however, is a good term

even if it implies artistic enhancement by corrosion, which is not always the

case. Patina as pointed out by Vernon and Whitby165 in their interesting

discussion of the term, is more than a tarnish. It connotes not merely a

specific color but also a continuous surface layer of measurable thickness.

In describing the reddish surface of some bronzes in the Freer, the late

John Ellerton Lodge, former Director, used the term "aerugo" (literally

copper rust); but this seems to have fallen into disuse. The term "verdigris"

sometimes used is entirely inappropriate since it specifically refers to the

green product formed by action of acetic acid on copper. For convenience

and because of its wide acceptance among collectors, we will often employthe term patina; but in purely technical discussions, we will speak as a

chemist of corrosion and corrosion products.

Probably all the Shang and Chou bronzes which have survived from an-

tiquity were either buried for centuries in direct contact with soil and ground

waters or were contained in the damp and little changing atmosphere of

hollow tombs. Conditions of burial are little known because most of the

bronzes were recovered from the earth in accidental finds or in clandestine

digging. Even professional archaeologists in modern excavations give little

attention to soil conditions or to the exact environment of burial. In recent

Chinese archaeological publications, however, there are photographic views

which show bronzes in the actual state of burial and first discovery.166 The165"The open-air corrosion of copper, Part II . .

.

," p. 407.166For example see Hui-hsien fa-chiieh pao-kao, plate 80, no. 1. Ho Han-nan in "Ch'ang-an P'u-t'u-t'sun . . .

," (K'ao-ku hsueh-pao 1957, no. 1, pp. 75-85) shows bronzes in situ in a Western Chou tomb (plate 1,

no. 6). Ch'en Meng-chia in Shou'hsien ts'ai hou mu cKu-Vu i-wu shows a large bronze hu as found in the tombof the Marquis of Ts'ai during 1956 (plate 2, no. 3). A hoard of several dozen bronze vessels apparentlyfreshly excavated from a late Chou dynasty tomb burial are illustrated by Carl W. Bishop in "The bronzes of

171

conditions of burial of Chinese bronzes are probably best summarized by

the late W. F. Collins, a British mining engineer who spent many years in

the vicinity of Peking. He says

:

"In China the deposit most frequently surrounding the bronzes is loess.

This is a yellow, fine-grained, highly porous, sandy clay, probably wind-

blown, containing much carbonate of lime. It is, geologically speaking,

about 60 m. in thickness, and covers practically the whole of the surface

of the Provinces of Chihli, Honan, Shansi, and Shensi, from which the

majority of Chinese bronzes are derived. Loess is a very favorable

matrix for the formation of patina. Through the historic period it has

held little humus on the surface, and oxidation under warm climatic con-

ditions has had free scope. Water, having a high content of carbonates

and oxygen, circulates through it with ease. The water table in the loess is

frequently, as in the great plain of North China, only a few feet from the

surface. The rainfall at a station north-east of Peking was 56 inches one

year, and about one-eighth of that amount the next. Alternate moistening

and drying of many of the bronzes may have been, over a number ofyears,

frequent. Water from the loess is very high in lime and other dissolved

solids, even at the surface. Shansi surface well-water contains from 295 to

299 mgr. per litre of solids. Water from a well in the loess at Taiyuanfu

contained 348 mgr. CaC03 per litre with gypsum and other sulphates

(Professor Nystrom). 1,167

Probably a large percentage of the bronzes buried in antiquity have

completely disintegrated and returned permanently to the soil. This is

especially true where the soil waters carried appreciable salt (sodium

chloride) content (fig. 234). A certain small number of buried bronzes,

however, because of favorable circumstances, have escaped total corrosion

and have survived to serve as precious documents of an ancient culture.

It is on these few survivals that we have to base our knowledge of the history

of metal technology in the Far East.

Of the three major metal constituents of the ceremonial vessels, the copper

in low-tin bronzes is probably the first element to be attacked ; this in spite

of the low position of copper in the electromotive series of elements. Onereason is that copper is the most abundant element in the alloy, usually

from three to four times as much as the combined amounts of tin and lead.

The second reason is that the common salts of copper, the chlorides, sul-

phates and carbonates are more soluble in soil waters than the correspond-

ing salts of tin and lead and hence are more easily carried away from the

Hsin Cheng Hsien." The vessels, which appear from the black-and-white photographs to be only moderatelycorroded, had been lying for about 25 centuries 1 2 to 1 4 feet below the surface in "yellow clay."167Collins, "The corrosion of . .

.," pp. 29-30.

172

Figure 228. Hu 11.47. A much enlarged detail from this

bronze shows small botryoidal malachite interspensed with

some lead corrosion products, a little cuprite and some rare

botryoidal azurite (macrograph, x 12.5).

Figure 229. Cross section of a specimen cut from a frag-

mentary Chou dynasty Chinese bronze hu, SC548 (Kelley).

The specimen was mounted edgewise in plastic, ground to a

flat surface, and then polished. The upper view taken of the

freshly polished specimen shows the layered structure of

copper and tin corrosion products that normally surround

remnants of the metal core. The lower view is the samespecimen after exposure to 100 percent relative humidity

for 24 hours. In two places along the inner zone next to the

metal the unstable mineral product nantokite (cuprous

chloride) has taken on water and oxidized to a newproduct called atacamite. This phenomenon is commonlycalled "bronze disease." x7.5.

Figure 228

174 Figure 229

Figure 230 Figure 231

Figure 232

Figure 230. Fang-i 30.54. The predominant corrosion pro-

duct in the vessel surface is cuprite or red cuprous oxide.

Figure 231. Fang-i 30.54. The inside of the lid in the area of

the inscription is covered with a smooth compact layer of

azurite; malachite surrounds this area.

Figure 233

Figure 232. Huo 36.6. The surface has a fine smooth patina

which is composed mostly of tin oxide stained green with

copper corrosion.

Figure 233. Chueh SC546. This unusual vessel made of

lead has a crusty patina of cerussite or lead carbonate.

175

metal surface. Fink and Polushkin, who studied many specimens of corro-

ded cast bronze, observed that in general metallic copper is first deposited

directly from the corroded a constituent; in the next stage the copper is

oxidized; corrosion of tin-rich a takes place simultaneously or later; and

finally the eutectoid is consumed. 168 In high lead and tin bronzes, the course

of corrosion is different; lead globules are early replaced by cuprite or trans-

formed to white lead carbonate (cerussite); the 8 phase in the eutectoid is

replaced by a dark corrosion product to leave islands of uncorroded a phase

quite unchanged in shape and position. The dark product, apparently high

in tin, is gradually transformed to white tin oxide. In the inner zones

attacked copper remains in the reduced form as cuprous oxide or metallic

copper, but in the outer layers it is oxidized to basic cupric salts like mala-

chite and atacamite.

The corrosion of metals is a complex phenomenon, which in spite of muchstudy and research is still not completely understood. Modern investigators

regard metal corrosion, especially under soil conditions, as an electro-

chemical phenomenon in which the electropositiveness or electronegative-

ness of metal plays an important part. These views are well summarized by

U. R. Evans in England who is one of the world's authorities on corrosion

of ferrous and non-ferrous metals. 169 There are numerous papers and mono-graphs on the subject because of the economic implications of corrosion.

This is no place, however, to attempt to review the various theories of the

mechanism of corrosion, especially when we know so little of the corroding

environment to which the bronzes were exposed. All we can do is describe

the end products which are found on the surface of those vessels which

have come down to us.

Many of these corrosion or alteration products are identical in compo-sition and form with the minerals that make up the ores of the earth's crust

from which the metal constituents of bronze were originally derived; they

are quite akin to the secondary minerals in the upper oxidized zones of

metal ore deposits. An earth mineral from which a metal was derived is a

compound of the metal which is in stable equilibrium with its environment.

The formation of corrosion products on a metal surface is the attempt of

the metal to revert to an identical or similar stable equilibrium. Because of

the similarity of corrosion products to well-known mineral species, we shall

use the terms of mineralogy to name them.

Malachite and azurite

Perhaps the most abundant alteration product on the vessels is one or the

168Fink and Polushkin, "Microscopic study . ..

,'"p. 103 ff.

169/1az introduction to metallic corrosion.

Ill

other of the two basic carbonates of copper,170 malachite and azurite; and as

the names imply, one is green, the other blue. They are formed by the action

of carbonated waters, either ground water or condensed water like dew or

sweat acting on the metal, and they are quite stable.

Malachite, CuC03-Cu(OH) 2 , occurs on bronze as it does in nature in

greater abundance than azurite. In some cases it is present as a bright green

continuous hard layer overlying an underlayer of red cuprite. Elsewhere,

as on the p'ou 13.30, it has the appearance of green enamel paint, but morecommonly it is found in encrusting patches which are variously described as

mammillary, botryoidal (like bunches of grapes), or tuberose (fig. 228). Mr.

Charles Muskavitch has called attention to a Han dynasty //w owned by Mrs. J.

C. Sebastian of San Simeon, California, which is so heavily encrusted with

mammillary malachite that it looks like a mineral specimen from a copper

mine {figs. 235 and 236).111 Internally the mammae or tubercles have the

fibrous radial structure with concentric banding that is so characteristic of

the massive malachite used for lapidary work. There is evidence that during

the initial attack on bronze, the copper that is taken into solution migrates

some distance along the surface before it is deposited as a malachite con-

cretion. Often on thinly gilded bronzes, the soluble copper salts pass in

solution through pores and breaks in the gold layer and are later deposited

as a crust over the gold. Occasionally, as on the side of the tsun 51.19,

malachite occurs on corroded surfaces in radiating crystals forming little

rosettes. According to Fink and Polushkin malachite is the end product of

copper corrosion, no matter what copper compounds were initially

formed. 172 This is probably true provided, of course, sufficient carbonic

acid gas is present to maintain equilibrium.

Azurite, the blue carbonate of copper, 2CuC03-Cu(OH) 2 , contains less

water and apparently forms under drier conditions than malachite since it is

often observed on the interior surfaces of vessels. On the inside of the cover of

ourfang-i 30.54 (fig. 231), it occurs as a thin compact and continuous coating

like paint enamel. More commonly it forms scattered patches of crystalline

blue on areas that are predominantly malachite. Rarely is it seen in botryoidal

form (see, however,^. 228). In copper ore deposits azurite frequently occurs

in well-formed indigo blue crystals, sometimes tabular, sometimes prismatic

in shape. Azurite can alter quite easily to malachite, but the reverse is rarely

true. In mineral specimens large crystals of azurite are sometimes seen170Several of the copper salts mentioned in this discussion are referred to as "basic" carbonates, chlorides, andsulfates. A basic salt as understood in chemistry is a salt composed of metallic cations, together with bothhydroxyl anions of a base and anions of an acid. Example: Cu2(OH)3Cl or as it is sometimes written in older

texts on mineralogy in double salt form, CuCl2-3Cu(OH>2. The corresponding normal salt is CuCl2. Basic

salts are usually less soluble in water than normal salts, hence occur more commonly in nature,l V iprivate communication.172Fink and Polushkin, "Microscopic study .

.

," p. 110.

178

Deep corrosion has caused a bronze chtieh in the Royal

Ontario Museum (960.234.90) to crumble and fall apart.

(Courtesy, Royal Ontario Museum, University of Toronto.

Photograph enlarged from a color transparency.)

Figure 238

P'an SC537. A view of a polished cross section shows that

nearly half of the alloy has been transformed to cuprite

(metallograph x 300, bright field).

Figure 236

Figures 235 and 236. A hu in the collection of Mrs. J. C.

Sebastian of San Simeon, California, is uniquely covered

with botryoidal malachite. The detail (fig. 236) shows

massive malachite covering the inside rim. Photographs

enlarged from color transparencies taken by Dr. Charles

Muskavitch of Auburn, California.

Figure 237

Chia 23.1. Cuprite often follows cracks deep below the

surface of the metal. Corrosion of this kind may account

for the low total for major constituents in the analysis of

samples from ancient objects (metallograph x 50, bright

field).

179

partially or completely altered pseudomorphically to malachite. This pheno-

menon may take place among bronze alteration products but is ordinarily

not easy to observe. Alternate bands of azurite and malachite, however, do

occur in a deeply corroded area on a bronze (M731-1910) in the Victoria

and Albert Museum; and they may be expected elsewhere.

Both azurite and malachite are important in the formation of noble

patinas on bronze, and they often contribute much to the appearance and

interest in an object and to its value in the eyes of collectors.

Cuprite

Hardly less abundant among alteration products on this series of vessels

is the lower oxide of copper, cuprite, CU2O, which is usually dark red in

color, but sometimes quite bright red. Cuprite appears to be an inter-

mediate product of basic salt formation and is found invariably under green

crusts of atacamite and brochantite (vide infra) but frequently underlying

malachite and azurite, as well (fig. 230). Plenderleith expresses the view that

cuprite represents invariably an early state in the mineralization of a buried

bronze and that its presence is an important criterion in establishing the

genuineness of its patina. 173It will be shown later, however, in the discussion

of basic copper chlorides that cuprite can be a secondary product produced

by the decomposition of initially formed cuprous chloride. Although in

polished sections of corroded metal cuprite is often seen apparently in

direct contact with uncorroded metal, it is probably produced as an inter-

mediate compound in the corrosion reactions (fig. 237). In some instances of

advanced corrosion, the metal core is completely transformed to cuprite with

little dimensional change in much the same way as in the pseudomorphic re-

placement of one mineral by another (fig. 238). In other instances of deep

bronze corrosion, however, cuprite seems to be entirely absent.

In polished cross sections of corroded bronzes, cuprite is sometimes

observed in the form of large rounded inclusions (fig. 239). They have the

size, shape, and distribution of lead inclusions in highly leaded bronzes,

hence they appear to be caused by corrosion of the lead and its replacement

by red cuprous oxide (fig. 240).

Most reddish crusts of cuprite are formless and massive, but often fresh

fractures at the surface of the red layer reveal a drusy aggregate of small

cubic or octahedral crystals. The color of a highly polished cuprite layer is

often bluish metallic.

Corroded bronzes in collections sometimes exhibit fairly large patches of

cuprite which have been uncovered by stripping agents applied by dealers.

173Plenderleith, "Technical notes on Chinese bronzes . ..

," p. 43.

180

Figure 239

Ku SC533. In the beginning stages of corrosion of bronze,

cuprite appears to outline the large polygonal grains of the

metal. The dark rounded pockets appear to be cuprite re-

placing lead (metallograph < 300, bright field).

Figure 242

Hu SC548 (Kelley bronze). The detail (x40, bright field)

shows a cross section of a line in the sunken decor. In the

mineral alteration the groove formed by the decor line has

been filled flush to the surface with compact red cuprite,

but the outline of the original groove is still plainly visible.

Figure 240

Ting SC613. The large rounded inclusions of cuprite seem

to be a replacement product for solid droplets of lead

originally distributed throughout the bronze alloy. Nearly

perfect cube-shaped crystals of cuprite are visible in the

large central inclusion (metallograph X 100, dark field).

Figure 241

Kuei 24.11. On the heavily corroded lid of this vessel someof the outer malachite and earth crust has cleaved away to

reveal original sunken decor well preserved in the under

layer which is mostly cuprite interspersed with tiny islands

of uncorroded bronze. The handle of the kuei is at the left.

(About 5 natural size.)

181

These agents are said to be fermented fruit juice or weak organic acids.

They dissolve the porous green crust readily but not so easily the hard

compact cuprite under layer. A dark red mass of cuprite on the handle of

kuei 60. 19 seems to have been formed in this way.

On many bronzes the thin compact layer of malachite or of mixed basic

salts does not attach itself strongly to the metal or to the cuprite under-

layer so that it cleaves away in flakes with moderate mechanical pressure

{fig. 241). This behavior often permits the successful mechanical cleaning

of bronzes with small hammers and chisels and percussion tools.

Cuprite infalse inlay

In the sunken decor on many bronzes, especially in areas of fine lei-wen, a

compact layer of red cuprite appears to line the bottom of the narrow

fossae, at the same time leaving the cloison-like ridges that separate them

green or grayish green. Some have mistaken the red deposit for paint or

inlay and have suggested that cinnabar was used for the purpose,

but this is not so. From one fragment of a broken and deeply corroded

Shang bronze vessel (SC548), it was possible to cut out, mount, and polish

a cross section of metal in a way that shows the cuprite-filled fossae in

cross section {fig. 242). The cuprite fills exactly the original channel. This

circumstance occurs again and again, but we can only speculate as to the

cause. It is known that polishing of metal has an inhibiting effect on its

corrosion because the softer constituents of the alloy smear over the

harder constituents and create a surface skin which is more homogeneous

than the unworked metal beneath. This smoothing over lessens the tendency

of localized impurities to form galvanic couples at the surface; or in words

of the electrochemist, it results in the "equipotentialization" of the surface,

thereby rendering the surface less susceptible to corrosion. The more pro-

tected surface of the fossae, however, were not smoothly polished; hence

they escaped equipotentialization and are corroded to form cuprite. This is

only speculation.

Basic copper chlorides

An unwelcome corrosion product on bronze is the basic chloride of

copper which occurs in two distinct crystal modifications, atacamite and its

dimorph, paratacamite, each having the formula CuCl2-3 Cu(OH)2. Ataca-

mite, often mixed with malachite, cuprite and other alteration products

forms ugly crusts and blister-like eruptions, sometimes uniformly, but at

other times only in local areas on bronze surfaces. When relatively pure,

atacamite occurs in small clusters or patches of dark green crystals, which

182

reflect light like granulated sugar. Sometimes, as on the side of p'ou 13.30,

it occurs in small cone-shaped formations, with stepped sides {fig. 243),

which indicates the corrosion effect is produced from an active center.

Atacamite crusts are not only unattractive, but they are associated with the

phenomenon called "bronze disease1' or "recurrent corrosion. " These

names are given to still active areas of corrosion which in humid weather

take on water and become moist, but in dry times of the year turn bright

green and slough off in fine granules. Although this phenomenon has

mystified museum curators and has given rise to wild speculation amongcollectors, it is fairly well understood technically, and a considerable amount

has been written about it. The well-known French chemist Berthelot seems

to have been the first to give a reasonable explanation of the phenomenonand later G. A. Rosenberg of Copenhagen also made significant contribu-

tions to our knowledge of "bronze disease/' 174 More recently R. M. Organ

has discussed with the aid of pictures and diagrams the theory underlying

various methods employed at the British Museum for stabilizing bronzes

afflicted with recurrent corrosion.175

There is in our study collection a fragmentary Chou dynasty hu known as

"the Kelley bronze" (SC548) which has been described in some detail in

Chapter V. The author some years ago described the internal structure of

the metal of this vessel and the corrosion phenomena including the attack

on copper by chloride agencies and the resulting products.176 Studies have

continued on this piece over the years. Polished metallographic sections

show a complicated structure. Along the irregular edges of the metal core

initial corrosion apparently first attacks and darkens the high-tin phase of

the duplex structure leaving the copper-rich a phase unattacked. This

natural etching clearly reveals the dendritic structure of the cast alloy

(fig. 244). Next to the remnants of the metal core is a gray, or pale green

soft wax-like layer of cuprous chloride (CuCl) identical with the unstable

mineral nantokite. This rather rare copper mineral is the source of bronze

disease. When exposed to humid atmosphere (lower limit about 50 percent

relative humidity), nantokite deliquesces and swells up as a thick greenish

liquid. The cuprous chloride not only hydrolizes but simultaneously oxi-

dizes to basic cupric chloride (figs. 245, 246 and 229). In a subsequent period

of low humidity excess water evaporates and leaves a powdery, bright green

residue which gives an X-ray diffraction pattern corresponding to parataca-

mite. Going outward the next mineral product is usually a thick layer of

bright red cuprous oxide or cuprite which may contain pockets, lenses or

174Berthelot, "Sur l'alteration . . . ;" Rosenberg, Antiquites enfer et en bronze, p. 50 ff.

175"Aspects of bronze patina . . .," p. 6 ff.

176"The corrosion products of an ancient Chinese bronze."

183

Figure 243

P'ou 13.30. A peculiarity of this bronze is the appearance

of conical protuberances with stepped sides that have

formed as the result of local corrosion in the vessel sur-

face. They seem to result from local galvanic action in-

duced by impurities or heterogeneous structures in the

vessel surface.

Figure 245

Figures 245 and 246. Hu SC548. Figure 245 (macrograph

x 3). Fragments of a chloride corroded bronze have been

mounted edgewise in plastic and ground and polished to

give a flat surface. Banded corrosion crusts line each side

of the bronze metal core. Figure 246 ( x 7.6, area outlined in

white on figure 245). After 24 hours' exposure to 100 per-

Figure 244

Hu SC548 (Kelley bronze). Photomicrograph (xlOO,bright field) of a polished section of metal. An uncorroded

remnant of the core is surrounded by a corrosion-attacked

zone in which islands of more resistant a bronze are sur-

rounded by dark corrosion product. This is in effect a

natural etching which reveals clearly the original crystal

structure of the alloy. In the lower part ghosts of tin oxide

formed from later corrosion of a dendrites are plainly

visible. (Courtesy, Daniel Cushing.)

Figure 246

cent relative humidity, pale green patches of oxidation

product, paratacamite, have erupted along the narrow

inner zone of nantokite (cuprous chloride) which lies next

to the metal core. When this happens on the exterior sur-

faces of ancient bronzes, it is called "bronze disease."

184

Figure 247

Hu SC548. Occasionally in the corrosion layers of ancient

bronzes, lenses of white tin oxide are abundantly distributed

in the dark-red cuprous oxide. This circumstance is shownin a mounted polished specimen from fragments of an

early Chou dynasty vessel ( x 100, dark field).

seams of white tin oxide (fig. 247). Between the nantokite and cuprite maybe a thin, irregular, and often discontinuous layer of pure redeposited

elemental copper. In other specimens crystals of pure metallic copper mayalso be scattered irregularly along the initial attack zone between nantokite

and the inner metal core. Beyond the cuprite and obviously formed from it

is the crusty green exterior layer which is the end product of copper altera-

tion. It may be nearly pure crystalline atacamite or a fine grained mixture of

atacamite and malachite.

Among the bronze vessels in this study only a few show evidence of

bronze disease. Perhaps the most seriously afflicted is the famous inscribed

fang-i 30.54 (fig. 248). Apparently when it came into the collection, it had to

some extent been mechanically cleaned to remove unsightly corrosion

crusts. The decapitation of blisters and excrescences uncovered pockets of

185

unstable cuprous chloride which later during damp summer seasons ab-

sorbed moisture and in the subsequent dry seasons were revealed as small

patches of powdery green (fig. 249). The attack has been arrested since the

installation of air conditioning in the galleries.

The extent to which Chinese bronzes in general are afflicted with bronze

disease is not exactly known. Perhaps this results from the scarcity of

scientifically excavated material. Badly corroded and disintegrated bronzes

found in clandestine diggings probably were thrown away at the site andonly those few found in comparatively good condition reached Western

markets. The author has seldom observed rampant bronze disease on Chi-

nese bronzes in American collections.

Tin-oxide patina

A number of Shang and Chou vessels have a smooth, gray-green patina

of almost eggshell quality sometimes called "water patina." This kind of

surface is found only on bronzes with high tin content and is mostly a thin

layer of hydrous tin oxide which results from the leaching out of the copper

constituent of the alloy and its replacement by the tin alteration product

without apparent volume change. The author has shown elsewhere that

early Chinese bronze mirrors, because of their especially high tin content

(20-25 percent), often furnish the best examples of water patina. The surface

alteration layer is sometimes loose and friable, and may even be powderywhite, but more often it is compact like a hard lacquer coating. Anygreenish color or undertone is a stain caused by residues of copper salts (fig.

232). The attacking agent which leaches out copper is probably ground water

charged with carbonic acid. The copper is taken into solution and trans-

ported away from the surface before it can be precipitated as basic carbonate

(malachite or azurite). In many instances, however, the basic carbonates are

deposited as a distinct surface layer which flakes or cleaves easily from the

tin-enriched surface (fig. 250).

In forming tin oxide the corrosion attack is generally along a solid front

parallel to the surface (fig. 251) and not intergranular as is usually the case

in corrosion fostered by chlorides (vide supra). Often, but not invariably,

cuprite is entirely absent. The reason for the separation of copper and tin

is the ready solubility of the initial copper corrosion products and the in-

solubility of the tin product; the tin oxide stays exactly where it is formed,

but the copper salts move away (fig. 252). On some vessels the tin oxide

layer is quite deep and forms a sort of shell all over the surface to a thickness

of a millimeter or more. It may scale off in small chips (fig. 253). In smaller

objects the metal may be entirely replaced by tin oxide with little or no

186

Figure 248

Figures 248 and 249. Fang-i 30.54. This object exhibits

some of the symptoms of "bronze disease." Figure 248.

The light area on the side of the top flange of the knob of

the lid is an area of active chloride corrosion. The pale

green powdery substance which causes the light area is

atacamite. Figure 249. The detail of the rim of the vessel

shows light colored areas of powdery atacamite (basic

copper chloride) surrounded by darker areas of red cu-

prite. The pale green areas are moist in extremely humidweather and powdery in dry weather. In the present air-

conditioned environment of the Freer Gallery the green

basic copper salts are stable (macrograph x 2.5).

Figure 250

Ting 46.31. On many bronzes that have been buried, the thin outer crust of malachite

cleaves easily from the bronze surface leaving a smooth pale green patina of tin oxide be-

neath. This circumstance permits cleaning by simple percussion methods.

187

Figure 251

Chia 23.1. View of the corrosion layers near the surface of a

bronze having medium tin content (ca. 14 percent). In this

kind of corrosion the a dendrites are converted to tin

oxide prior to attack on the eutectoid (metallograph

> 100, bright field).

Figure 252

Pou 13.30. The surface is mostly covered with a uniform

and smooth dark green patina composed of tin oxide. On a

mounted and polished cross section of this bronze viewed

in bright field at x30 (unetched) the tin oxide is the

crinkled zone (about 0.5 mm. thick) between the dark

plastic mounting material (upper) and the bright core

metal. The upper edge of the tin oxide follows exactly the

original surface of the metal; the line of penetration into

the core is comparatively straight and uniform. In high-tin

bronzes the copper that is dissolved away in the corrosion

process is replaced by tin oxide with little or no volume

change or distortion.

Tsun 25.2. The pseudomorphic replacement of copper by Figure 254

tin (stannic) oxide causes little apparent alteration of the Chia 23.1. Copper corrosion products formed below the

original smooth metallic surface. Often the tin oxide re- surface are the cause of blister-like formations in the tin-

placement is brittle and chips away from exposed edges. oxide patina.

188

distortion or swelling. Bronze mirrors will even retain their polished surface

although deeply corroded. The tin oxide layer, however, is not always con-

tinuous; it sometimes is raised from underneath to form blisters which in

turn have a smooth uninterrupted surface (fig. 254) and as previously men-

tioned on bronzes that are deeply encrusted and distorted by chloride

corrosion, the cuprite zone is often interlaced with pockets or lenses of tin

oxide like seams of fat in raw beefsteak (fig. 247).

Tin oxide is insoluble in nitric acid, but sufficiently soluble in dilute

hydrochloric acid to yield a microchemical test (with cesium chloride) for

tin. The product gives an X-ray diffraction pattern characteristic of cassi-

terite, SnCh, although the lines are more diffuse than in the natural mineral

(fig. 255). The diffuseness of the lines indicates that the particle size is small,

which explains why the product appears nearly amorphous when viewed

microscopically in transmitted light. The refractive index for the tin oxide

alteration product is in the range n= 1.78 to 1.81, much lower than the value

for cassiterite which is definitely crystalline (tetragonal, w=1.99, €=2.09).

It is known also that the tin oxide formed by corrosion loses some water onheating. It probably has a composition approximating metastannic acid,

SnCVEhO. 177 H. Otto (1961) has proposed for this material the term hydro-

cassiterite and he also calls attention to a relationship between tin oxide

formed on bronzes and a fairly recently discovered tin oxide mineral called

"varlamoffite,,,

which occurs in the Belgian Congo and in the ancient tin

mines of Cornwall.178 Like tin oxide corrosion product, varlamoffite gives

the X-ray diffraction pattern of cassiterite but with diffuse lines. They seem

to be identical compounds (fig. 255). The author has treated this topic morefully elsewhere.179

The pseudomorphic replacement of the surface of high-tin bronze with

tin oxide is not peculiar to Chinese bronzes. It has been observed abundantly

on bronzes of the late Bronze Age in Europe. The writer has observed it onhigh-tin bronzes from Yugoslavia in the Duchess of Mecklenberg Collection

at the Peabody Museum at Harvard University, and he has seen some beauti-

ful examples on the surfaces of Etruscan bronze mirrors. The most complete

study of pseudomorphic replacement of copper in bronze by tin oxide wasmade by Professor Geilmann of Mainz who has shown by many quantita-

tive analyses that the tin oxide contains a certain proportion of combinedwater; hence, he prefers to call the product by its chemical name stannic

acid (Zinnsaure). 180 This relationship between tin oxide corrosion product

177tTin oxide patina . .."

178"t)ber rontgenographish . . .," p. 663.

179Gettens, "The corrosion products of metal antiquities," p. 561.180"Verwitterung von Bronzen . .

.," p. 201

.

189

Cassiterite (Sn02)

Araca, Bolivia

USNM R8034Film number F191

Metastannic Acid

(Sn dissolved in HNO3;precipitate air dried.)

Film number Fl 319

Varlamoffite

Bunny MineSt. Austell, Cornwall

USNM R8886Film number F1311

Tin oxide patina

from p'an

FGA 56.26

Film number F251

Tin oxide patina

from elephant-shaped huo

FGA 36.6

Film number F364

Tin oxide patina

from /;;/

FGA 59.14

Film number F1057

Figure 255

The sharp lines of the X-ray diffraction pattern of tin oxide

mineral, cassiterite, contrast strongly with the diffuse lines

of metastannic acid, varlamoffite, and tin oxide corrosion

products from ancient bronzes.

Figure 256

Kuei 11.58. The underside is thickly encrusted with a

whitish deposit of cerussite (lead carbonate) derived fromthe lead constituent of the alloy.

Figure 257

Lid of yu 16.361. Much of the surface is deeply corroded

and mineralized, but about a third of the total area is

nearly as smooth and lustrous as when it left the makers'

hands. Bald areas like this on ancient corroded bronzes are

not too unusual.

190

from ancient bronzes, the mineral varlamoffite, and stannic acid seems to be

confirmed by the X-ray diffraction patterns of the three materials made in

the Freer Laboratory (fig. 255).

Corrosion products oflead

Lead, which is usually the lesser of the major constituents of bronze, also

corrodes. On a number of our second quality vessels, especially among those

that are heavily corroded, grayish crusts are found which consist largely of

cerussite or lead carbonate, PbC03 {fig. 233). This product, because it is also

dense and white, may not be easy to differentiate from tin oxide when both

are present but it occurs more commonly than is usually realized. Quite

perfect crystals occur on the underside of a kuei 11.58 {fig. 256) and ting

11.60. Often cerussite is mixed with malachite, and at times it is tinged red-

dish with litharge (lead monoxide). Occasionally cerussite forms a distinct

layer beneath crusts of malachite. When seen in long-wavelength ultraviolet

light, cerussite fluoresces pink; hence, these areas may be mistaken for

evidence of paint or repair. Sometimes in polished cross-sections of corroded

bronze little round pockets of a white deposit can be seen surrounded by

matrix metal; these were originally unalloyed droplets of lead which have

been converted to cerussite.

Other elements

With perhaps one exception, the minor and trace elements in bronzes

seem to have little or no effect on the color and nature of the patina. Thesmooth green patina of bronze yu 47.12 has a slightly yellowish tone in

contrast to the bluish tone of other vessels in the collection. It is noteworthy

that this is the one vessel in the collection which has iron content above

1 percent (actually 2.3 percent). Apparently enough iron oxide is present

to give the patina a warm tone.

Bald (unaltered) areas on corroded bronzes

One sees now and then an ancient bronze on which large areas are deeply

corroded and mineralized, but where adjacent areas on the same piece are

little affected and retain their original brown metallic and lustrous surface.

A good example of this kind of corrosion resistance is seen on the lone cover

of a ceremonial vessel, probably a yu 16.361 (fig. 257), where there are

several square inches of unaltered bronze surrounded by thick green and red

corrosion crusts. The bald areas show no sign even of first stages of corros-

ion beyond a thin tarnish. Another example is a hu 24A2(fig. 258). Here an

191

Figure 258

Hu 24.12. A bald area about 10 cm. in diameter on this otherwise moderately corroded vessel was found

concealed under imitation patina. After removal of the falsification the original surface was revealed as

shiny yellow metal, which for some reason had escaped corrosion.

uncorroded yellowish metallic area about 10 cm. in diameter was found con-

cealed under modern false patina in imitation of the surrounding thin

natural corrosion crusts. Probably some former owner felt the bald area

made the object look too new, so he hid it. There is an interesting example of

isolated patches of lustrous surface surrounded by crusty corrosion on the

lid of a yu in the Fogg Art Museum (1943.52.120), Harvard University;

and the surface of another, a yu shown in the catalogue (p. 11) of the

Hakutsuru Museum, Kobe, Japan, miraculously seems to have escaped

with only small patches of rough patina. The highly polished surfaces of

Chinese bronze mirrors often show thick patches of crusty green alternating

with areas of bright metallic lustre. The reason for passive areas is hard to

explain. Plenderleith, who has observed this phenomenon suggests that

isolated uncorroded areas are purer and more noble than the corroded

areas in the galvanic sense and hence are protected from galvanic action.181

Fossil impressions oforganic materials

There is one other interesting aspect of corrosion and that is the imprints

or impressions left by organic materials on bronzes that were long buried.

Apparently vessels were often wrapped in cloth or other fabric because on181"Technical notes on Chinese bronzes . . .

," pp. 35-36.

192

the surface of a number of them small areas of weave are plainly visible.

Contact of fabric with the bronze also may have been accidentally made at

the time of collapse of a tomb. In the green crusts on the inside of the yu37.1 {fig. 259), a clear imprint of twill-woven straw matting occurs; and there

is a distinct imprint of a finely woven fabric, perhaps silk, on the inside

surface of a tsun 51.19 {fig. 260). Sometimes the fossil impressions of fabric

are faint and difficult to see without good light and some magnification. Onsome of the nipples that decorate the side of a ting 59.15 {fig. 261), a fine

weave pattern shows in the tin-oxide surface. Fossil fabric impressions also

show up on other ancient buried objects, mirrors, weapons, and even onnon-bronze objects like jade. Usually the organic material has been entirely

replaced by copper minerals, but occasionally residues of the original still

remain. These fossil impressions give some idea of the history of a bronze

and its burial environment, and they may be important factors in judging

the age and authenticity of a piece. Vivi Sylwan of Stockholm was able to

determine from the fabric imprints on a bronze axe from An-yang that a

twill technique of weaving was employed in Shang times. 182

Conclusion

Although corrosion products do bear evidence of the past history of a

bronze, they are not thoroughly reliable in establishing age and authenticity.

Time is only one of the several factors that determine the kind and extent of

corrosion; the nature of the corroding agencies and the kind of environ-

ment including temperature, availability of moisture and oxygen, depth of

burial, alternation of wet and dry seasons, and many others all contribute.

Bronze patina has many origins hence interpretation of its meaning must be

made with care. Some years ago Yetts stated the situation very well:

"Corrosion is an even more complex problem than that of metallic com-position. Here also there is need for full knowledge of contributory fac-

tors. Who can correlate cause and effect without full data of environ-

ment? Captain [William F.] Collins gives some useful tables showing the

composition of loess in different localities, and the fact is well known that

most of the territory of ancient China was covered with loess. But this

does not help us much to assess the influence of purely local conditions,

such as the proximity of organic matter and associated objects in the

place of burial. And there remain, besides, incalculable factors, such as

duration of burial, and the vicissitudes through which a bronze may have

passed above ground. In short, a reasonable conclusion seems to be that,

in our present state of knowledge, patina is but a minor guide as to the

182"Silk from the Yin dynasty."

193

Figure 259

Fw, 37.1. The impression of some woven vegetable

material, probably straw matting, is retained in the

corrosion layer at the edge of the vessel.

Figure 260

Tsun 51.19. Detail showing a patch of corrosion on the

vessel interior in which an imprint of a woven fabric is

retained. It is perhaps a ghost impression of the fabric used

to wrap or lay over the bronze at time of burial (macro-

graph « 2).

Figure 261

Ting 59.15. The imprint of a woven fabric is retained on three of the knobs which are part of the vessel

decor. The ting was perhaps wrapped in a cloth before burial.

194

age of a bronze. Certainly, the absence of patina does not preclude

high antiquity." 183

83"Problems . ..

p. 401.

CHAPTER IX

FILLINGS, INLAYS AND INCISED DECOR

Although Shang and even early Chou ceremonial weapons were often

sumptuously decorated with inlay of turquoise and sometimes malachite,184

the ceremonial vessels were not. Vessels of those early times were decorated

only with cast-in designs. It was not until late Chou and well into Han that

semi-precious stones were employed in vessel decor, and it was during these

times that gold and silver inlay came into wide use, not only for decorating

vessels, but also for ornamenting weapons, harness fittings, and objects for

personal use. J. G. Andersson has shown how the use of gold and silver

marks a new era of style and technology in metal working in China starting

around the sixth-fifth centuries B.C. 185 A few of the Freer vessels are richly

decorated with precious metal inlays, but before treating these we should

consider inlays of less noble materials.

Blackfillings

In many vessels of Shang and early Chou it is noted that some black

substance fills neatly and precisely the sunken decor (figs. 262 and 263).

Quite often the black appears to be a deliberate filling which during burial

became encrusted and covered over with mineral alteration product

(fig. 264). H. G. Creel speaks of the black and calls it a pigment, and he

pictures a Shang vessel of the type chueh which is typical. 186

Much effort on our part has been expended in an attempt to identify the

composition of the black substance, but with only partial success. It wassuspected at first that the black might be oxide of copper (tenorite) or copper

sulphide (chalcocite or covellite), but there is no supporting evidence for

any of these. Nor is it niello, that artificial black substance so often used in

antiquity for ornamenting silver. Quite early in the studies microsamples

of the black were taken for X-ray diffraction analysis. Usually the powderpattern showed that the material is predominantly quartz, or quartz mixedwith cuprite (fig. 265). At first the quartz was regarded as a casual impurity;

but microscope studies showed some curious features. Quartz particles makeup the bulk of the black; but instead of being rounded particles of assorted

sizes, the grains are uniform in size and sharply angular (fig. 269). The184Some of these are illustrated in Lodge, Wenley and Pope, . . . catalogue ofChinese bronzes.185"The goldsmith. . .

."

186 77k? birth of China, p. 1 13 and plate VI.

197

Figure 262

Ting 47.11. The black filling in the fossae of the design is made up mostly of finely divided quartz and car-

bonaceous material. The black seems to be an intentional part of the decor (macrograph x 3).

Figure 263

Fang-ting 50.7. Again the filling is a mixture mainly of carbonaceous material and quartz (macrograph x 2).

198

Figure 264

Ting (privately owned, V54.60). The black filling in the deeply sunken lines of the decor appears to be

intentional inlay. In various places on this vessel, the black is overlaid with green mineral corrosion

products. (Courtesy, J. T. Tai, New York.)

Quartz

standard

White in fossae

Bronze ku

FGA 43.9.

Quartz

Black "inlay"

Bronze ting

FGA 47.11

Quartz, mainconstituent

Red "inlay"

Bronze kuang

FGA 38.5

Cuprite, mainconstituent

plus quartz

#

•Figure 265

X-ray diffraction patterns of specimens of black filling of the sunken decor of ancient bronzes show that

quartz is the most common inorganic component. Quite often cuprite is also present as a corrosion

product.

199

quartz component is quite similar in particle shape and size distribution to

the quartz particles already described in Chapter IV in connection with

earthen cores which seem to be made from ordinary loess. It is recalled that

loess is made up 60-80 percent of quartz particles; and in spite of its sup-

posed aeolian origin, the particles are sharp and broken and remarkably

uniform in size.

It was further observed that when a particle of the black is warmed with a

drop of concentrated nitric acid, the black disperses and on drying leaves a

ring of brown amorphous material. It has all the appearance of carbona-

ceous matter which is true of specimens of black in a number of vessels.

There has never been any evidence of fine opaque even-sized particles

characteristic of lamp-black pigment which forms the basis of Chinese ink.

These observations confirm, in part, those made independently by H. J.

Plenderleith who notes: "The black powdery inlays in Shang bronzes have

proved to be mostly residues ofcarbon with some silica and phosphorous. This

seems to support the common belief that such black inlays in Shang bronzes

may be remains of some primitive form of lacquer of organic origin." 187

As Plenderleith suggests, the black filling may indeed be residues of a

lacquer filling. The late Bishop William C. White, in his description of

the contents of Han tombs at Lo-yang, mentions (1934, p. 25) a large ting

covered with a geometrical pattern that had been incised and filled with

black lacquer. He says further: "The treatment of lacquer inlay is not

unusual for vessels of the San Tai period. Bronze ritual vessels that would

probably date back to Shang or at latest early Chou, very often are found

with a deep inlay of black lacquer in the decorative design."

He speaks of others and he concludes: "As far as one's observation goes

this treatment was characteristic only of certain types of ritual vessels,

particularly the ting and the chioh, while certain other types were not treated

in this way."188

Microchemical tests on the filling from two vessels, huo 33.2 and ting

47.11, and others, show that in addition to carbonaceous material and

quartz, the black filling tests for lead ion, and X-ray diffraction analysis

shows that the lead is present as lead carbonate (cerussite). It is not knownwhether the lead is part of the black filling or whether it comes from cor-

rosion of the bronze underneath, but it could be the latter.189

Some of the most interesting examples of black inlay have been seen in

pieces outside the Freer collection. There is a broken fragment of a ku in

187"Technical notes on Chinese bronzes . ..," p. 38.

1887bw/w ofoldLo-yang, p. 25.189Bishop White (Bronze culture . . . , p. 147) also speaks of a mixture of lime and pig's blood which tradition

says was used in early China to make an inlay similar to lacquer. No indication of lime (calcium carbonate) wasfound in the black fill of Freer bronzes.

200

Figure 269

Figures 266-269. Ku SC529. The black filling in the sun-

ken decor of this fragment of a Shang dynasty beaker

could be studied in detail. Figure 266. The filling, obviously

intentional, was applied perhaps for the purpose of

heightening contrast of the design (about xl.5 natural

size). Figures 267 and 268. A whitish filling in the sunken

decor is pure fine white quartz ( x 8). In a polished cross

section from the same area (fig. 268) the quartz filling in a

fossa is seen in end view. The space between the quartz andthe original sides of the groove is filled with black carbon-

aceous material. The next light area going outward is

cuprite, and the outer dark area shades into partially

corroded metal (x40, dark field). Figure 269. Photo-

micrograph (x270, crossed Polaroids) showing angular

particles of quartz in the filling.

201

our study collection (SC529, fig. 266), which shows such a filling in the

fossae. Since it was a fragment, it was possible to cut out a specimen of the

metal bearing the filled design and study it in metallographic polished

cross-section (figs. 267 and 268). It shows that the fossae are packed with

fine white angular quartz particles and that the black is lacking except for

a thin layer on the surface. It is certainly not loess fill. Also, a sort of white

filling was observed in the fossae of another ku 40.3 (fig. 270). This also is

mainly finely powdered quartz (figs. 271 and 272).

A different kind of black filling was found in the sunken decor of a ting

in the Royal Ontario Museum (ROM 947.33.1) (fig. 273). Here the car-

bonaceous material is mixed with filings of metallic copper (figs. 274 and

275). The particles are so well and evenly dispersed that there is little doubt

that they were deliberately added to the black. 190

There is another puzzling situation. On several of the vessels which have

sunken decor with black filling, the black extends beyond the sunken decor

and covers the in-between areas. On some bronzes large areas are covered

with black. A fang-ting 50.7 (fig. 276) and a p'an 56.26 (fig. 277) are exam-

ples. Specimens of black from these areas also contain carbonaceous mater-

ial and scattered particles of quartz. In most instances of this kind it is im-

possible to differentiate between black from the fossae and black on mi-

decorated surfaces. This has led the writer at times to regard the black

deposits in sunken decor as evidence that the vessels were actually employed

as cooking utensils.191 During generations of use over the hearth, wind-

blown earth (loess) could have been trapped in the fossae by oily fumes and

soot in much the way it does in the corners and crevices of the modern bake

tin or frying pan. Both quartz and carbonaceous material are as enduring

as bronze. Occasionally one finds in the sunken decor, as in a chiieh 56.19

(fig. 278), accretions of loess soil which is not mixed with black. This contains

the same high proportion of angular quartz particles as the black accretions.

If black deposits originated in such fashion, after burial they could becomemixed with and overlaid by mineral corrosion products; and centuries later,

when the mineral crusts are cleaned or chipped away mechanically, the

black crusts are revealed lodged in fossae of the design and giving the ap-

pearance of filling or inlay. Whether the black is accretion or intentional

filling is sometimes hard to tell.

To make the problem even more difficult, former owners in some in-

stances have artfully touched in the sunken decor with black paint in

190O. Karlbeck (private communication) has suggested that lines and lei-wen with vertical or even undercutsides were intended to hold fillings or inlay.1910. Karlbeck (private communication) says there is a 4th-3rd century B.C. ting in the Ethnological Museum,Stockholm, which carries in its bottom part of the skull of a bird embedded in the residues; apparently a left-

over from the last feast.

202

Figure 270 Figure 271 Figure 272

Figures 270-272. Ku 40.3. The filling in the sunken decor

of some vessels is light gray or even whitish in tone.

Figure 270. Here the fossae of the fine lei-wen are partially

filled with whitish transluscent brittle material which seems

originally to have been applied as a paste. Microscopic and

X-ray diffraction analysis shows that the substance is

mainly powdered quartz ( x 6.5). Figures 271 and 272. In a

Figure 273

Figures 273-275. Ting in the Royal Ontario Museum(947.33.1). The wide spaces in the heavier sunken decor of

some of the Chou dynasty bronzes sometimes carried a

filling or inlay. Figure 273. Embedded in the residues of

black inlay of this vessel are fine irregularly-shaped bits

crushed sample viewed microscopically by transmitted

light (fig. 271) the filling can be seen as a mixture of trans-

parent and opaque particles of fairly uniform size. Theopaque particles are mostly carbonaceous material. Thetransparent particles seen between crossed Polaroids

(fig. 272) are mostly of quartz ( x 170, mounted in mediumofn= 1.65).

Figure 275

of metallic copper. (Courtesy, Royal Ontario Museum,University of Toronto.) Figures 274 and 275. The bits of

copper seem to be filings and they are rather evenly distri-

buted in a matrix of some carbonaceous material (photo-

micrographs 120).

203

imitation of black filling. This was done on a bronze ting 59.15 and on the

kuang 49.10 {fig. 279). It is appreciated by the dealers that a black-filled

decor is more readable than an unfilled one. Painted black decor, however,

can easily be recognized under a magnifying glass (see also Chapter X).

About the only other accretions on the vessels are residues of earth from

burial that are not completely cleaned off but are lodged in corners and in

the sunken decor. Often it is entrapped or mixed with corrosion products

just like the black filling. It is probably good policy not to regard these

earthy accretions as common dirt and clean them away, but to allow traces

of them to remain, because they may bear evidence of the kind of place

where the bronze was buried.

Inlays with semi-precious stones and metals

As mentioned above, the decorating of vessels with jewels and inlays of

precious metals did not become common in China until the sixth century

B.C. or later, a view supported by the few inlaid vessels in the collection

which are ascribed to late Chou or Han or even to more recent periods.

Unfortunately, there are too few of these in the Freer to serve as a basis for

even guarded generalizations. A few comments, however, may be ventured.

The bird decorations of a single vessel, a rather crudely made hu 24.12,

are enriched with jewels of turquoise. The stones were perhaps held in their

sockets by a cement because the edges of the sockets seem not to serve as

bezels. Turquoise and silver wire were also used to decorate bronze tun

38.7 in a rather unique way. Terminal areas of the linear silver design were

deeply and precisely recessed and the recesses filled with tiny turquoise

tesserae {fig. 280). Only a few scattered fragments of the turquoise remain,

chiefly in the medallion on the center of the lid. In its original condition

the vessel must have been splendid. Jewels of malachite are set in the

handle of i 61.31. Malachite inlay is used much more lavishly and in a

different way in hu 61.32 {fig 281) assigned to the late Chou dynasty and in

tui 11.81 of more recent origin. In these vessels the chips of banded mala-

chite are set in grooves as in a mosaic, and are probably held in place by a

cement. Toward the very end of Chou inlays of other semi-precious stones,

including carnelian and jade, come more and more into use, especially on

gilded objects.

The metals employed for inlay are copper, silver and gold. Copper is

used in only two vessels in the series and both are ascribed to late Chou. In

the handles of hu 51.22 parallel square strands of copper looped back at the

ends are set deep in recesses of sunken decor. In hu 61.32 copper inlay is

combined with malachite mosaic. The copper is looped back and forth six

204

Figure 276

Fang-ting 50.7. The sides and underside of this vessel are

partially covered with a carbonaceous crust; evidence per-

haps that the vessels were actually used to cook food.

Figure 277

P'an 56.26. A black carbonaceous coating not only fills

sunken decor but also covers broad surfaces between the

decor lines. This kind of black appears to be an accretion

of smoke and earthy residues. (Detail, about § natural size.)

Figure 278 Figure 279

Chiieh 56.19. Macrograph (x2.3) showing Kuang49A0. The detail (macrograph x 2.9) shows how the sunken

earthy residues in the sunken decor partially lines of the decor are touched up and strengthened with black

covered with black. paint.

205

Figure 280

Tun 38.7. The sides and cover are lavishly inlaid with silver

now heavily tarnished. The inlay on the recessed rim of the

vessel is freshly cleaned. The hollows were formerly filled

with bits of turquoise, now mostly fallen out.

Figure 282

Tou 39.41. The inlay is made of strips of gold of varying

width from 0.5 to 2.0 mm. and about 0.2 mm. thick, set

into grooves. It adheres well in spite of the heavy corrosion

of the surface (macrograph / 1 .3).

Figure 281

Fang-hu 61.32 (lid). This large vessel is elaborately decor-

ated all over with inlay of malachite and copper metal set

into cast-in cloison-like dividers. The chips of malachite

show the banding characteristic of that mineral. Copperstrips twisted back and forth are hammered into alternate

sunken areas to give the impression of solid copper. Thesquared ends of the strips are visible in a slightly damagedarea near the bottom of the illustration (macrograph x 2.4).

Figure 283

Pien-hu 15.103. Detail (before cleaning) showing the silver

inlay partially concealed by copper corrosion products.

206

Figure 284

Lien 46.11. The decor of running animals is deeply chased

into the surface of the metal.

Figure 286

Lien 51.5. The relief decor in the side of the vessel appears

to have been cast, but the fine line detail was cut or chased

in after casting. This is the kind of decor sometimes found

on vessels in late Chou or early Han periods.

Lien 51.2. Detail showing chased decor. The running

animal figure bridges the gap between lid and vessel for

proper orientation.

Figure 287

Chien 15.107. The tiny figures in the hunting scenes on the

vessel have the appearance of being etched in the surface by

some acid or other corroding material but they were

probably cast in.

207

or seven times to fill a broad channel with parallel strands. Time has tar-

nished the copper to a brown tone that is hardly distinguishable frombronze.

Silver and/or gold inlay is not well represented since it occurs on only

four of our late Chou vessels. The delicate silver inlay in tun 38.7 {fig. 280)

already mentioned, consists of narrow strands of squarish cross-section set

in channels precisely cut into the bronze surface. The black tarnished

silver now makes a pleasing contrast with the gray-green patina of the

bronze. The broader strips of gold inlay in ton 39.41 {fig. 282) appear to be

cut-outs from gold sheet which are also set into cleanly and squarely cut

recesses in the surface. The silver inlay that almost completely covers pien-hu

15.103 {fig. 283) is made of still broader cut-outs in a style familiar in

decorated objects of the late Chou and Han periods. It is quite remarkable

how well the inlay is preserved on this hu and on other vessels that are quite

deeply corroded. The plastic qualities of silver and gold apparently per-

mitted the inlay to be hammered tightly against the walls of the grooves andfriction still holds them there.

There were other ways to decorate bronze surfaces. Lien 23.2 of the Hanperiod is traced with a fine line decor on the outside and gilded all over

probably by the fire gilding process. The underside of the cover bears a

delicately painted design of birds and animals against a red-painted back-

ground (see Vol. I).

Other lien of this same shape and period bear designs that are tooled

in the surface. On those shown in figures 284 and 285 the outlines are sculp-

tured and the fine detail is engraved, and in the lines of the latter the chatter

marks of the engraver's tool are plainly visible. In another lien {fig. 286) the

low relief animal decor is probably cast, but it has a sculptured quality; all

of the fine detail, however, the checkered background and the diamond

pattern appear to be chased. The delicate sunken figures that decorate the

sides of a chien {fig. 287) have an acid etched quality, but since mold join

marks run vertically through some of the decor elements they were probably

cast in. The hollows at one time may have held silver or gold inlay, but if so

no trace remains.

208

CHAPTER X

FALSE PATINA AND REPAIRS

There has been much speculation about false patina on Chinese bronzes

but little factual information. Barnard devotes a whole chapter to it192 in

which he brings together bits of information on true and false patina from

widely scattered modern technical reports, but he also quotes passages

from the Chinese literature of Sung and Ming that deal with the faking of

bronzes by inducing corrosion artificially. Although some of the old recipes

read like passages in writings of medieval alchemy they may have some

basis of truth. They include a portion of the often quoted passage by KaoLien (A.D. 1591) from Tsun sheng pa Men (Ch. 14, p. 28) "Hsin chu wei

tsao" (Newly cast forgeries). 193 Certainly treatment with corrosive salts,

burial in dung heaps, heating and sprinkling with vinegar, pickling in fer-

menting fruit juices, and other manipulations will induce a patina, and all

these tricks are known today in modern art bronze foundries. In laboratory

examination of our bronzes, however, we find only slight evidence of this

kind of falsification— but it is possible that the products of induced corro-

sion are so similar, chemically and visually, to those of natural corrosion,

that we cannot recognize them. On the other hand, we have seen quite a lot

of falsification and repair; about one-third of the vessels in the series showmodern repairs ranging from scattered touches of paint to replacement of

major parts.

The finding of paint retouches on bronzes that have passed through a

long line of previous owners is not unusual. The purpose of paint is at least

three fold: First, to conceal repairs; second, to cover up dull and ugly

patches of cuprite exposed by partial mechanical cleaning; and last, to

liven up and prettify the patina. The location of paint is usually quickly

revealed under ultraviolet light as a pale pinkish fluorescence especially if

there is oil or resin in the paint medium; but where the paint vehicle is

animal or vegetable glue, there may be a little or no fluorescence. Paint is

revealed in other ways. Under pressure of a needle it is usually softer and

more plastic than natural corrosion products, and it is not so adherent.

Where thick paint is spread over sunken decor, air trapped in the grooves

192Bronze casting . . . , Chapter 7, pp. 198-217, Study of patina and corrosive effect.193See Jung Keng, Shang chou . . . , pp. 195-196; see also Barnard, op. cit., pp. 204-213. Dr. Jan Fontein ofBoston has called attention (private communication) to a passage in Chin-shih-lu by Chao Ming-ch'eng(1081-1 129), ed. of 1762, ch. 11, p. 4A which indicates that forgeries were known in Northern Sung.

209

rises as bubbles to the surface and leaves tiny rounded pits or craters

{figs. 292 and 293). Crusty and porous imitation patina is friable and under

pressure usually cleaves easily from the surface, whereas natural patina is

hard and gritty under the probe. The most convincing evidence of paint andfalse patina is the identification of paint pigments especially those known to

be modern.

MATERIALS OF FALSE PATINA

1. Ground mineral malachite and azurite. These natural pigments madefrom the copper mineral are still used by artists in the Far East and are

available to repairmen and fakers. Since these pigments are identical in

composition with natural copper corrosion products, they cannot be dis-

tinguished by chemical means. They are fairly easily recognized micro-

scopically, however, by the random orientation of the crushed mineral

particles (fig. 290, color).

2. Emerald green or Paris green, which is an artificially-made compound(copper aceto-arsenite) was, according to the best information available,

first produced in Schweinfurt, Germany in 1814. In our experience Paris

green is the most commonly used pigment in touch-up paint on bronzes.

Being a copper compound and having the proper bluish undertone, it serves

admirably for imitating natural malachite and atacamite. Fortunately,

Paris green is easily recognized under the microscope by its particle

characteristics. When fragments of paint are dispersed in a mounting

medium like Canada balsam on a microscope slide and observed in strong

transmitted light at magnifications above x 400 the particles of Paris green

are seen as bright emerald green spherulites more or less uniform in size.

The tiny green discs usually have a dark center (figs. 288 and 289, color).

Microchemical tests on an additional small sample of paint can be made for

both copper and arsenic; a positive reaction for the latter is strong confir-

mation of Paris green.

3. Prussian blue. This artificial pigment, which was invented in Europe

at the beginning of the 18th century, is a sort of dyestuff and hence stain-like

in character, but because of its true blue tone it is often used to imitate

natural azurite patina. It can be easily identified in small paint samples by

simple microchemical methods. The color in a tiny paint particle is de-

stroyed when it is immersed in a 5 percent solution of caustic soda (sodium

hydroxide), but the color returns when the test drop is acidified, provided

there is a high concentration of blue in the sample. Presence of Prussian blue

is indicated when a cotton swab wet with alcohol or other strong solvent is

stained blue when rubbed on the surface of a bronze. Natural patina does

not stain the swab.

210

4. Artificial ultramarine, invented in Europe and known commercially

since about 1 830, is occasionally identified in paint samples from trumped-up

bronzes. It too is easily recognized microscopically and can be confirmed by

microchemical tests on small paint samples. The blue color is quickly

destroyed by dilute hydrochloric acid; and if enough of the pigment is

present, the odor of hydrogen sulfide can be detected simultaneously.194

5. Smalt is a sort of powdered glass which gets its color from the element

cobalt. It has been used for centuries as a paint pigment in the Orient as

well as in the West. In concealing repairs it is useful where a coarse texture

paint is wanted. It is insoluble in acids and alkalis but can be recognized

microscopically (figs. 294 and 295).

6. Chinese writing ink (J. sumi) from lamp black is used occasionally to

strengthen or falsify the black inlay in sunken decor as described in Chapter

IX (fig. 279).

7. Several other modern pigments including zinc oxide, barite, cobalt

blue, and iron oxide have been detected in false patinas. They can be

recognized by standard methods of pigment analysis.

Not so much is known about the mediums or vehicles of these false

patinas. Often a thick paint is used so it can be applied as a sort of daub to

imitate the texture as well as the color of corrosion crusts. Starch grains

have been seen in samples of paint from various bronzes {figs. 294 and 295);

This vehicle is used where it is desired to imitate a crusty surface. Sometimes

odors that suggest shellac or natural resins are noted when tiny specimens

of paint are heated on a platinum spoon. Animal glue is a useful mediumbecause, even when used in small quantities, it is strongly adherent and it

does not fluoresce under ultraviolet light. Some paints are so remarkably

resistant to organic solvents that one is led to suspect that Oriental lacquer

(J. urushi) was used.

MENDS AND REPAIRS

Many bronzes that come through commercial channels are broken and

mended with soft lead-tin solder. Sometimes missing parts like handles andknobs or even broken-out pieces are replaced by soldered-on new metal

parts which are concealed with cleverly-made artificial patina. Some of

these mends have held fast but others, because lead solder does not adhere

well to old corroded metal, are weak, and they have often come apart in

transit or from rough handling.

It is not known when soft solder came into use in China, but if we can

rely on the information supplied by Kao Lien cited above, it was used in

194See Joyce Plesters, ".. . Ultramarine Blue . . .

."

211

Ming times. None has been found on any of the authentic bronze vessels of

Shang and Chou in the Freer collection except in connection with modernrepairs. This is strange because both lead and tin, the two important in-

gredients of soft solder, were known to Chinese metal craftsmen of the

Classical Periods. H. Maryon says that soft solder has a very long history;

and he cites a Sumerian copper relief panel of before 3000 B.C. in which

soft solder is used.195It is possible that, even though tin-lead alloys were

known, they could not be used for joining metals because of lack of know-

ledge of fluxes; but fluxes of some kind must have been used for brazing.

Fluxes are substances used to treat the surfaces of heated metals to remove

or hinder formation of metal oxides which prevent molten solder from wetting

the surfaces to be joined. They may be made of inorganic salts like borax or

zinc chloride or organic substances like rosin or tallow. Since no informa-

tion was available on the composition of the soft solders used to repair

Chinese bronzes, analyses were made on samples taken from vessels in

the Freer collection. The compositions listed in Table XII are fairly uniform

and quite close to the composition of modern plumber's solder.

TABLE XII

Analysis of soft solders used in modern repairs*

Object Cu%

Sn

%Pb%

Total

%

Bronze hu SC 548(Kelley) 50.6 43.9 94.5fBronze p"an SC 537 (Kriger) 0.4 52.6 45.8 98.8

Bronze ku SC 542 3.2 46.9 47.2 97.3fHandle of bronze chien FGA 15.107 0.9 58.0 39.4 98.3

Cylindrical jar FGA 04.256J 1.7 54.4 42.7 98.8

*Wet methods of analysis used.

fSpectrometric analysis shows Sb> 1 %. Analysis incomplete

JNot included in catalogue.

Around the areas of solder repair on most old bronzes there is little or no

blackening of the original patina or other evidence that high heat was used

in making the repair, and this suggests that solders of complex composition

and hence very low melting points were used. Tin-lead 50:50 solder melts at

220°C; five parts tin to three of lead melts at 180°C. These melting points

195Maryon and Plenderleith, "Fine metal-work," p. 653; this early use of soft solder is borne out by state-

ments in J. R. Partington, Origins and development of applied chemistry, p. 64, and by A. Lucas, AncientEgyptian materials . . . (1962), p. 253 ff. The Romans used soft solder freely to join metals (Mantell, Tin . . . ,

p. 318) and the Elder Pliny described (Historia naturalis, Book XXXIV, Par. 160) two tin-lead compositionsthen in common use: "Tertiarium" consisting of tin-lead in the ratio one to two, and "Argentarium," tin andlead in equal ratio.

212

Figure 288A. The particles of Paris green (copper aceto-

arsenite), an artificial pigment often used by the Chinese in

making imitation patina, are easily recognized because of

their rounded shape, even size and dark centers. This

sample from a commercial source was photographed by

transmitted light at > 675.

Figure 288B. The spherulitic structure of radiating crystals

in each particle is shown in the same view taken between

crossed Polaroids.

Figure 289A. Particles of Paris green were recognized in a

sample of artificial patina on p'ou 09.334. The green is

mixed with fine particles of colorless inert pigment. Viewby transmitted light at x675.

Figure 289B. Same field as "C" taken between crossed

Polaroids.

Figure 289A Figure 289B

Figure 290 Figure 291

Figure 290. Artificial patina is sometimes made by strew-

ing coarse particles of crushed malachite mineral on a

vessel surface made adherent with glue. Many of the ran-

domly oriented particles show the typical banding of

massive malachite, x 20.

Figure 291. Ku SC542. This forgery, purchased in China,

was constructed around five fragments of the foot of anancient and genuine vessel. The entire upper portion is

fabricated (in sections) from a modern bronze containing

about 5 percent zinc. Some of the relief design is made in

lead. The new metal surface was given a false painted

patina now partially removed.

215

Figure 292

The artificial nature of the patina on a bronze i (privately owned) is

indicated by the many small craters and pits caused by the escape of

air bubbles during application of a thick, viscous coating used to

simulate natural patina (macrograph x 5.5).

Figure 293

The artificial patina on a bronze hu

(privately owned) is porous and crusty, a

condition which was caused by entrapment

of air in the pasty mixture when it wasdaubed on and its subsequent escape

during film drying (macrograph X 1.7).

Figure 294 Figure 295

Figures 294 and 295. Tsun 51.19. False patinas often contain curious mixtures of pigments. This sample

of false patina from the surface of one of our vessels contains Paris green, smalt, Prussian blue, andcolorless mineral filler. It also contains starch grains which served as the adhesive. These grains show as

dark crosses when the same field is viewed between crossed Polaroids {fig. 295). (Photomicrographs

X400.)

217

Figure 296 Figure 297

Figures 296 and 297. Chio 53.83. Some repairs were madeby skillful artisans. Figure 296. Two legs of this vessel

were broken off and rejoined with soft lead solder but the

join is well concealed by paint which closely imitates the

color and texture of the natural blue and green copper

corrosion products. The areas of repair, however, are

easily revealed in ultraviolet light. Figure 297. The radio-

graph shows the exact location of the breaks in the two

legs and the solder used to mend them.

218

Figure 298 Figure 299

Cu Ka

26 i

degrees 10

Figure 300

Figures 298-300. Hu 13.21. Each pair of horns of the two

animal head handles of this vessel is false. Figure 298. Aview of one handle shows how well the repairs are con-

cealed with false patina. Figure 299. The radiograph

reveals the presence of two metal dowels used to secure the

replacement horns to the handle. Figure 300 (chart). Thedifference in composition between original metal of the

vessel and replacement metal in the horn of the animal

head handle is shown by X-ray fluorescence analysis. In

the trace of the original metal (lower) a considerable

amount of tin and some lead is shown. In the trace of the

repair metal (upper) the alloy element is mostly zinc.

(Courtesy, Conservation-Analytical Laboratory, Smith-

sonian Institution.)

219

Figure 301 Figure 302

Figures 301 and 302. Kuang 49.10. This vessel has been in both original and falsified parts has been retouched with

rather extensively damaged, but skillfully repaired. Figure black paint. Figure 302. In this radiograph (taken without

301 . The false handle made of zinc bronze is concealed with lid) the solder joins along the rim of the vessel as well as

a painted false patina which fluoresces more strongly in the solder joins of the new handle to the stumps of the

ultraviolet light than the vessel proper, and hence appears original handle are revealed,

light in the photograph. The filling in the sunken decor

Figure 303

Ting 47.20. A photograph taken of the fluorescence exhibited when the bottom of the

ting was exposed to ultraviolet light shows false patina (light areas) where the third

mold mark should be. (Compare withfig. 32.)

220

Figure 304

Figure 305

Figures 304-306. Chien 15.107. The vessel has been rather

extensively altered. Figure 304. None of the handles is

original but are probably recent additions. Three of themare old, but the fourth is an imitation. The detail shows the

soft solder join at the base of one handle and also the

colored plaster which was used to conceal it. Figure 305.

Figure 306

Radiographs show that the handles are secured to round

bosses cast integrally with the vessel. Figure 306. Theradiograph taken in profile shows the two bosses in the

vessel side which serve to anchor one of the handles, and

also the solder that secures them.

221

Figure 307 Figure 308

Figures 307 and 308. Hu SC548 (Kelley). Fragments of

this broken Chou dynasty bronze were joined (in China)

with soft solder and the joins and replacement parts con-

cealed with colored plaster to make a seemingly whole andintact vessel. During transport to the United States the

object came apart at the joins revealing its fraudulent

Figure 309

Figures 309 and 310. Hu SC548. Macrographs ( x 8) were

taken of details of the decor on the surface of the vessel.

Figure 309 shows the natural corroded surface of a single

unit of lei-wen decor on the old and original part of the

nature. In figure 307, taken in ultraviolet light, the simu-

lated patina fluoresces, but the areas of genuine patina

along the rim do not. In figure 308 taken in ordinary light

the good matching qualities of the natural and artificial

patina are shown.

Figure 310

piece. Figure 310 is a detail of the same feature of the

design as shown in figure 309, but imitated in plaster and

tinted and toned to the dull green color of the natural

surface.

222

may be further lowered by addition of antimony, bismuth, cadmium or

zinc, but with decrease in strength.

Solder joins concealed by paint and false patina are easily revealed by

probing with a dental tool or a heavy dissecting needle ground to a chisel

point. The softness and plasticity of the solder and its leaden color are un-

mistakable. They also show plainly in radiographs, because of their high

lead content.

The extent of modern repairs on ancient pieces varies widely. Some are

simple mends of members that broke off presumably when excavated or

from careless handling. Seen in ultraviolet light chio 53.83 (fig. 296) fluores-

ces where two legs join the body, and probing at the joins revealed soft sold-

er. The legs appear to be original. The breaks as shown by X-rays are clean

and the solder is thin (fig. 297): the repairs are concealed with a highly con-

vincing false patina which causes the fluorescence. Fang-i 15.136 has lost a

portion of its foot; it was repaired with a soldered-in piece of scrap metal

from another old bronze and the repair concealed with paint. The rim ofp'ou

13.30 and a portion of the sidewall of kuei 11.53 were patched in the same

way. The ears of both animal heads on the handles of hu 13.21 are

replacements made of brass; X-rays (figs. 298-300) show they are secured

to the heads with soft solder and metal dowels. Quite often new metal is

used in repairs. The handle of kuang 49.10 which looks genuine enough is

in fact shaped from a modern piece of brass. The handle metal contains

about 20 percent of zinc, whereas the metal of the vessel proper contains

none. The joining, which is made with soft solder, is well concealed

with paint which fluoresces moderately in ultraviolet light (fig. 301). Aradiograph (fig. 302) shows extensive repairs. Missing portions of a large

ting 47.20 (figs. 32, 223-225 and 303) have been replaced with cut-outs of

sheet copper which are held in place with soft solder and the mends con-

cealed with artificial patina which also fluoresces in ultraviolet light.

Losses and extensive repairs on vessels are sometimes concealed with

plaster and paste compositions. The tips of two of the legs of chueh 25.3 are

made with gypsum plaster (plaster of Paris) and the plaster is painted to

match the natural patina. On the square ting 09.261 an area of loss in an end

wall has been filled with plaster and painted with a pigment mixture in

which Paris green, smalt and ground azurite were identified. On hu 48.1 an

old crack which extends from the rim to one of the lug handles has been

luted with plaster and the mend concealed with bright green paint which

again contains Paris green. Much colored plaster has been used to conceal

the solder joins of the replacement handles fixed to the large chien 15.107

(figs. 304-306).

223

The replacement pieces and the joins on the aforementioned "Kelley

bronze" (Chapters Vand VII) were hidden under a green-tone plaster com-position carved to imitate sunken decor and colored and pock-marked to

simulate natural copper and tin corrosion products (figs. 307-310). Theplaster varies from 1-3 mm. in thickness; it adheres well to the old frag-

ments, but poorly to the new smooth copper. The plaster is softer and moreeasily scratched with a knife than the natural smooth gray-green tin-oxide

patina. It is little affected by water but is disintegrated and detached from

the bronze surface when immersed in alcohol. The alcohol extracts a resi-

nous binder which is stained green with copper. Identification of this binder

was not certain, but a sample when burned gave off an odor reminiscent of

shellac. The main component of the plaster left after resin extraction is a

white powder tinted green in which spectroscopic analysis shows the

presence of large amounts of zinc and copper and lesser amounts of silicon,

aluminum, barium and arsenic; sulfate and acetate radicals were also

detected. The X-ray diffraction pattern of the plaster shows the presence of

barite, and microscopic examination reveals bright green spherulitic crystals

of Paris green, which explains the presence of the copper, arsenic and acetate

ions previously detected. The object was truly a clever forgery until collapse

of the patched-together assembly revealed the flummery; but it has proved

useful because the old and genuine parts have yielded much valuable study

material.

A somewhat similar stained plaster was used to conceal repairs to the

bottom of the large ting (SC587) already mentioned in Chapter IV. In this

instance, the upper portion and the three legs had probably been found in

some excavation or grave robbery, but the bottom was broken out and lost.

Undaunted the forgers supplied a new bottom of sheet copper in which holes

were cut to allow insertion and attachment of the legs. The entire replaced

and mended areas were covered with crusty plaster made to imitate the tone

and texture of the authentic parts. Again, under stress the mends gave way and

revealed the fraud ( fig. 311). Another interesting piece is a ku (SC542) ; the foot

of the vessel is made up of five genuine old fragments, and the single charac-

ter inscription on the inside of one is evidently Shang. The remainder of the

vessel, however, is made from cast brass and the whole bears a false patina

which very cleverly imitates the water patina of an early high tin bronze

(fig. 291, color). The false patina is quite insoluble in strong organic solvents

as well as in water which suggests that the medium may be Oriental lacquer.

One more much battered vessel, already referred to, is a p'an (SC537) which

bears a 10-character inscription. It is said that this vessel known as the "LoPo p'an" arrived in the United States from China in broken condition. One

224

entire side had collapsed and examination of the fragments showed that

they were genuine, and part of the original p'an (SC537) {fig. 312). Again the

mends were made with soft solder and the repairs concealed with false patina.

A strange looking fang-i in the Smithsonian Institution Division of

Ethnology (fig. 313) is an obvious forgery made of heavy thick flat plates of

brass skillfully soldered together and covered with a uniform false green

patina that simulates malachite remarkably well. Microscopic examination

shows that it is made of uniformly sized pale bluish green particles, but

obviously not malachite and not a commercial paint pigment. X-ray

diffraction analysis shows that the pigment corresponds to the mineral

paratacamite, which often occurs in the natural patina of ancient bronzes

but not in this form. This false patina might have been induced directly onthe bronze by one of the recipes described in the Sung and Ming passages

referred to above but it seems more likely, from the uniformity of the

coating and particle character of the pigment, that the green was produced

by corroding copper filings or scraps in an aerated brine solution, andafter drying, was ground fine to make a paint pigment. The coating seems

to have an organic vehicle, but it is insoluble in strong organic solvents

and again it may be Oriental lacquer.

225

Figure 312

A pl

an (SC537) known as the Lo Po p'an arrived in the United States in fragments. Examination

showed that it is a genuine piece which was broken in many pieces, probably in an accident. In

China the pieces were joined with soft solder and all repairs concealed with false patina. The frag-

ments have served well for technical studies.

Figure 313 Figure 314 Figure 315

This fang-i is a forgery in every

respect. In addition to being a Figures 314 and 315. Hu 59.14. Figure 314 shows the condition of the

crude imitation of a Shang dynas- vessel when acquired in 1959. The pale green corrosion layer whichty vessel in type and decor, each seems quite stable is composed mainly of hydrous tin oxide. Figureside of both vessel and lid is cast 315, which shows a polished dark surface, was taken many years agoseparately of copper-zinc alloy when the object was in a private collection. Apparently the vessel in

(brass) and the parts joined to- recent years was given an artificial patina to make it more saleable,

gether with soft solder. The green

surface is false patina.

226

It is not known exactly where the manufacture of bronze vessels from odd

fragments and scraps has been carried on, but certainly most of it was done

in China. Peking before the war was probably the center for such activities

because a large demand for antiquities existed there. Some such center was

needed to supply the cheap yet skilled labor required for the work. It is

possible that some of the repairs and falsifications are very old and such

vessels were intended not for foreign trade but for the local market.

There is some speculation and talk about bronzes in the antique market

that are said to have been given an appearance of age by pickling in fer-

mented fruit juices or other corroding agent. This may very well be true in

certain instances but direct and convincing evidence of this kind of treatment

is scarce. There are several vessels in the Freer with more or less uniform

frosty, pale green patina, which is suspected of being recent in origin. Aphotograph of one of them, hu 59.14, taken when it belonged in a previous

collection, shows it with a black patina. This object seems to have under-

gone "improvement" in a New York dealer's shop. It now has a fairly

uniform pale green patina, mostly tin oxide {figs. 314 and 315). One suspects

that the other vessels with similar surface may have been "pickled" in this

country to make them more attractive to collectors. Many visitors to Taiwanhave observed that the bronzes formerly in the Imperial Collection of

Peking and now housed in the National Palace Museum in Taipei have dark

metallic surfaces and little of pale green patina. The presumption is that the

bulk of these have been above ground for centuries and have been cleaned

and polished during that time by their Chinese owners. On the other hand

most of the bronzes that have found their way into Western collections

have been unearthed, either accidentally or by clandestine digging, in the

present century and have retained their corroded surfaces to one extent or

the other. Once the Western taste for Chinese bronzes was established,

it was the bright green surface, whether smooth or crusty, that was in de-

mand; and as occasional dark green or dark brown bronzes with polished

surfaces became available from old Chinese collections, the dealers were

only too ready to meet the European and American demand by artificial

means.

227

APPENDIX I

THE VALIDITY OF THE ANALYTICAL DATA

During the course of the chemical analytical investigation of the Freer

bronze ceremonial vessels, a number of questions arose with respect to the

correctness and validity of analytical data published by previous investi-

gators; and in turn, questions arose about the accuracy and significance of

our own data. We realized that analytical chemistry is not an exact science

but an art based on scientific principles, and for this reason the results are

no better than the skill of the analyst. It was necessary to decide the

accuracy needed in order to answer properly the questions raised. If a figure

on metal content of an alloy that is accurate within 1 percent will answer

the question, it is not necessary to strive for a figure accurate to within 1 part

in ten thousand. The higher the accuracy, the greater the time required and

the greater the cost of the analysis. In choosing the analytical procedure,

attention had to be given to methods of sampling, homogeneity of alloy

structure, effects of corrosion, and other factors that would affect validity

of the results. In order to test some of these factors, a few side experiments

were carried out, and tests were made to demonstrate the validity of the data.

Constancy ofcomposition

During the investigation questions arose about constancy of composition

in respect to different parts of the same bronze. It was first desirable to

know if a single bronze member or cast part was homogeneous throughout

or if composition showed variation at the extremities of a piece. In

Table A are shown the comparative compositions of a limited number of

samples taken from widely separated locations, usually the rim and the foot

of a vessel. In all instances the analyses show small but definite differences in

composition which are felt to be beyond the limits of experimental error.

The chemical analyses tend to confirm metallographic and radiographic

observations that the alloy of many vessels is not entirely homogeneous,

probably because of the irregular distribution of the lead, or the uneven

penetration of corrosion. They both point out the difficulty of getting a

truly fair sample from an ancient bronze and offer reasons why duplicate

analyses on samples from adjacent areas frequently do not quite agree. It

also raises a question of the validity of second-place and especially of third-

place decimal figures that are so often casually reported by analysts and con-

fidently accepted by uncritical readers.

229

TABLE AConstancy of composition

Comparison of analyses of different parts of the same vessel member

FGA Location of Cu Sn Pb Fe* TotalISO. 0/

/o0//o

0//o

0//o

0//o

09.260| foot 79.7 14.0 1.2 94.9rim 82.8 14.1 1.8 98.7

11.35 vessel (foot) 71 .6 9.2 15.0 95.8

p'an handle 69.0 7.8 20.1 96.9

1 1 1113.21 toot

OA 180.7 10.5

1 13.2 94.4

hu rim 80.8 10.8 3.9 95.5

i r* 1135.12 under handle 75.0 15.6 9.5 100.1

chia capped column 76.8 14.3 8.8 99.9

38.5 foot 69.1 15.1 13.9 98.1

kuang rim 68.5 16.2 13.4 98.2

40.11 foot 81.4 13.6 3.6 98.6rim 82.8 14.6 3.7 101.1

47.12f foot 74.8 13.1 3.7 91.6rim 74.7 12.6 5.4 2.1 94.8

Wet analyses of duplicate, approximately 100 mg. samples.* Blank space indicates element not sought for.

tThese vessels are obviously too corroded to yield fair samples for analysis.

Composition ofmembers ofa multiple-member assembly

Many of the bronze vessels have separately cast covers and handles.

A number that appear to be cast in one piece are not. Careful examination

has shown that in many bronzes the legs, handles and appendages were

cast separately and joined to the vessel proper by casting-on or by other

joining techniques. It might be expected that the various vessel memberswere cast from the same melt, and inspection of the analyses in Table Bshow that in certain instances this is the case; but in others it is not. Com-ment on the groups of analyses shown in Table B appear in the "Technical

Observations" of the catalogue entries for individual vessels (Vol. I). Theresults show that the analyst must be careful to specify the member or place

on a bronze from which the sample for analysis is taken.

The accuracy andprecision ofanalyses done by wet chemical methods

Most reports on analyses of ancient bronzes make little or no mention of

230

TABLE BMultiple member vessels

Comparison of analyses of different member parts of the same vessel

ruANo.

Location ofsample

Cu%

Sn/o

rb%

re*

%"N T I *

0//o

J otal

%

07.37 bottom 82.3 12.2 1.0 — 95.5cli inL f 1 114 bird on nnstI'll V^l V ' 1 1 I / V ' , 1

1

83 4 14.8 4 98.6

leg (cast-on tip) 93.6 3.3 — — 96^9

15.136 rim 67.8 7.1 21.9 — — 96.8

fang-i cover 67 .7 7.3 22.2 — 97.2

24.11 rim 65.8 7.2 22.8 — — 95.8

kuei handle 76.4 8.0 12.4 — 96.8

ininint* mpta 1pV'IIIIIB i—, 111V ICll 60.4 22.8 9.4 92.7

30.26 base 67.4 14.3 13.4 — 95.1

yu 70 4 1 3 7 1 2 8JL \J 96.9

appendage 68.8 13.1 14.0 — — 95.9

handle 73.5 12.4 11.6 — 97.5

30.54 rim 77.7 21.6 1.2 — — 100.5

fang-i cover 77.9 20.3 0.9 — 99.1

31.10 base 70.3 10.0 14.7 — — 95.0

kuei handle 70.7 10.7 13.7 — — 95.1

38.20 foot 82.0 13.2 3.6 98.8

kuei handle 82.3 14.4 2.3 — — 99.0

39.5 base 70.8 12.8 14.7 — 98.3

chien handle (cast on) 69.2 14.3 14.2 trace 97.7

40.11 foot 81.4 13.6 3.6 — — 98.6

yu cover 82.5ICC1 5.5 2.0 yy.v

handle (swing type) 89.0 10.1 1.2 100.3

AIM foot 74.8 13.1 3.7 2.3 trace 93.9

yu cover 75.9 13.8 4.6 2.1 96.4

handle (swing type) 77.4 14.0 6.1 97.5

47.20 rim 74.8 13.7 10.0 98.5

ting leg 73.6 11.7 11.4 96.7

Wet analyses of duplicate, approximately 100 mg. samples.

*Blank space indicates element not sought for.

231

the size of the sample employed, the methods used, or the precision (repro-

ducibility) and the accuracy (correctness) claimed.196

The difficulties that arise from this neglect are twofold: first, without

access to information bearing on sample taking, sample size and analytical

method employed it is not possible to make critical comparison of the

analyses made in different laboratories; second, non-chemists who quote

or otherwise employ other people's results are inclined to be non-critical

and to accept the figures given by the analyst as exact and accurate even

when they are reported to the third decimal place from a single analysis. Aspreviously intimated, non-chemists often fail to realize that analytical

chemistry is to a considerable degree an art and that many factors affect the

reliability of the results. Neglect to comment on accuracy and precision of

analysis is a common failing of nearly all analytical chemists who have

dealt with archaeological materials. Dr. Willard F. Libby, who developed

the radiocarbon method of dating organic materials, rendered outstanding

service to archaeological chemistry by insisting that all radiocarbon dates

be reported in a manner to show probable error in the form of plus or

minus limits. It is difficult to state precisely the probable error in any wet or

spectrometric analysis, but it is essential that the analyst in some way in-

dicate for the benefit of his non-technical audience how good his analytical

findings are, and for the benefit of professional colleagues enough details

about sampling and methods of analysis employed so that the professional

can make his own judgments. For everyone's benefit all analysed objects

should be identified by reference to a photograph.

In this study, accuracy was tested in two ways: first, the analytical method

was proved by analyzing a metal sample of known composition.

The known or standard sample was one of a series of standard

samples that is available from the U.S. National Bureau of Standards. Theresults of a series of analyses on this standard are shown in Table C.

Although better accuracy for copper and zinc is attained with a 1.0 gram

sample than with a 0.1 gram sample better accuracy for tin and lead was

actually attained on the smaller samples. This may be explained by the fact

that the smaller samples were analyzed at a much later time when the analyst

196T. Dono, however, in the analyses he reported over a quarter century ago gives weight of sample anddescribed in outline the analytical method employed. (The chemical investigation of the ancient metallic

culture in the Orient, p. 310 ff.) The analytical data on Chinese bronze mirrors tabulated by the Japanesechemists Komatsu and Yamanouchi in 1933 are given to two decimal figures, hence, are presumed to beprecise and accurate, but there are no details of method. (A chemical study . . .

, p. 295 ff.) Shu-chuan Liangand Kan-non Chang in their more recent report on analysis of Chinese Bronze Age materials (The chemicalcomposition . . . , p. 9 ff.) briefly outline their method of wet analysis and speak of following standard analytical

procedures; but they give no precise information on sampling; nor do they mention size of the sample. Themore recent Japanese investigator, Giichi Tanabe has, however, taken pains to a commendable degree to

describe his methods of sampling, the amount of sample taken, and the details of both the wet chemical andthe spectrometric methods he employed. (A study on the chemical compositions . . . , p. 271 ff.)

232

TABLE CAccuracy (correctness) of analyses

Comparison with National Bureau of Standards alloy,

Standard Sample 124c, Ounce Metal.

Expressed in percentages, showing 95% confidence limits

Cu%

Sn

%Pb%

Zn%

Total

%

NBS analysis 84.22 (±0.01*) 5.13 (±0.03*) 4.74 (±0.04*) 4.93 (±0.03*) 99.02

FGA analyses

Ci(1.0g.1

sample)ay

84.05

84.01

84.03 (±0.25f)

5.50

5.49

5.50(±0.06t)

4.35

4.57

4.46 (± 1.40|)

5.02

4.98

5.00(±0.25f) 98.99

C(0.1g. 1sample)

ay

83.49

82.90

83.20 (±3.78f)

5.03

5.24

5.14 (± 1.34f)

4.90

4.77

4.84(±0.83f)

3.52

4.05

3.79(±3.37f) 96.97

C3 (0.1g. 1sample)

ay

84.01

84 0984.05 (±0.50f)

5.22

4.91

5.06 (± 1.96f)

4.87

4.79

4i83(±0.50f)

—

93.94

C 4 (0.1g. 1sample)

ay

84.03

83.68

83.96 (±2.22|)

4.86

5.05

4.96 (± 1.21f)

4.70

5.02

4.86(±2.04f) 93.78

Average of

C2, C3 and C483.70 (±0.72$) 5.05 (± 0.051) 4.84(±0.04J)

*Based on 4 determinations,

fBased on 2 determinations.

JBased on 6 determinations.

had achieved greater skill through practice. Also, the semi-micro-analytical

set-up was better adapted for 0.1 gram samples. It was fortunate, because

of the limitation in size of sample that could be taken and because of

the non-homogeneity of the alloys, that these analyses of Freer bronzes did

not demand a high degree of accuracy.

Accuracy was also tested by comparing the results obtained in one

laboratory with those obtained on the same material by analysts working

independently in other laboratories. Fortunately, some data are available